Package 'Colossus' reference manual

Title:	"Risk Model Regression and Analysis with Complex Non-Linear Models"
Description:	Performs risk analysis using general non-linear models. Risk models can be the sum or product of terms. Each term is the product of exponential/linear functions of covariates. Additionally sub-terms can be defined as a sum of exponential, linear threshold, and step functions. Cox Proportional hazards <https://en.wikipedia.org/wiki/Proportional_hazards_model>, Poisson <https://en.wikipedia.org/wiki/Poisson_regression>, and Fine-Gray competing risks <https://www.publichealth.columbia.edu/research/population-health-methods/competing-risk-analysis> regression are supported. This work was sponsored by NASA Grants 80NSSC19M0161 and 80NSSC23M0129 through a subcontract from the National Council on Radiation Protection and Measurements (NCRP). The computing for this project was performed on the Beocat Research Cluster at Kansas State University, which is funded in part by NSF grants CNS-1006860, EPS-1006860, EPS-0919443, ACI-1440548, CHE-1726332, and NIH P20GM113109.
Authors:	Eric King-Giunta [aut, cre] (ORCID: <https://orcid.org/0000-0002-1577-766X>), Amir Bahadori [ctb] (ORCID: <https://orcid.org/0000-0002-4589-105X>), Dan Andresen [ctb] (ORCID: <https://orcid.org/0000-0003-2345-6695>), Linda Walsh [ctb] (ORCID: <https://orcid.org/0000-0001-7399-9191>), Benjamin French [ctb] (ORCID: <https://orcid.org/0000-0001-9265-5378>), Lawrence Dauer [ctb] (ORCID: <https://orcid.org/0000-0002-5629-8462>), John Boice Jr [ctb] (ORCID: <https://orcid.org/0000-0002-8755-1299>), Kansas State University [cph], NASA [fnd], NCRP [fnd], NRC [fnd]
Maintainer:	Eric King-Giunta <[email protected]>
License:	GPL (>= 3)
Version:	1.6.4
Built:	2026-10-15 23:21:54 UTC
Source:	https://github.com/ericgiunta/colossus

Fully runs a case-control regression model, returning the model and results

Description

CaseControlRun uses a formula, data.table, and list of controls to prepare and run a Colossus matched case-control regression function

Usage

CaseControlRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  conditional_threshold = 50,
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)
CaseControlRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  conditional_threshold = 50,
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

control

list of parameters controlling the convergence, see the control options vignette for details

conditional_threshold

threshold above which unconditional logistic regression is used to calculate likelihoods in a matched group

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

norm

methods used to normalize the covariates. Default is 'null' for no normalization. Other options include 'max' to normalize by the absolute maximum and 'mean' to normalize by the mean

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- CaseCon_Strata(Cancer_Status, e) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CaseControlRun(formula, df,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4)),
  control = control
)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- CaseCon_Strata(Cancer_Status, e) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CaseControlRun(formula, df,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4)),
  control = control
)

Fully runs a cox or fine-gray regression model, returning the model and results

Description

CoxRun uses a formula, data.table, and list of controls to prepare and run a Colossus cox or fine-gray regression function

Usage

CoxRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)
CoxRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

norm

methods used to normalize the covariates. Default is 'null' for no normalization. Other options include 'max' to normalize by the absolute maximum and 'mean' to normalize by the mean

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4)),
  control = control
)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4)),
  control = control
)

Fully runs a cox or fine-gray regression model with multiple column realizations, returning the model and results

Description

CoxRunMulti uses a formula, data.table, and list of controls to prepare and run a Colossus cox or fine-gray regression function

Usage

CoxRunMulti(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = matrix(c("temp00", "temp01", "temp10", "temp11"), nrow = 2),
  realization_index = c("temp0", "temp1"),
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  fma = TRUE,
  mcml = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)
CoxRunMulti(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = matrix(c("temp00", "temp01", "temp10", "temp11"), nrow = 2),
  realization_index = c("temp0", "temp1"),
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  fma = TRUE,
  mcml = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

realization_columns

used for multi-realization regressions. Matrix of column names with rows for each column with realizations, columns for each realization

realization_index

used for multi-realization regressions. Vector of column names, one for each column with realizations. Each name should be used in the "names" variable in the equation definition

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

fma

a boolean to denote that the Frequentist Model Averaging method should be used

mcml

a boolean to denote that the Monte Carlo Maximum Likelihood method should be used

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- floor(runif(nrow(df), min = 0, max = 5))
df$rand0 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand1 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand2 <- floor(runif(nrow(df), min = 0, max = 5))
realization_columns <- matrix(c("rand0", "rand1", "rand2"), nrow = 1)
realization_index <- c("rand")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(t0, t1, lung) ~ loglinear(dose, rand, 0) + multiplicative()
res <- CoxRunMulti(formula, df,
  control = control,
  realization_columns = realization_columns,
  realization_index = realization_index
)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- floor(runif(nrow(df), min = 0, max = 5))
df$rand0 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand1 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand2 <- floor(runif(nrow(df), min = 0, max = 5))
realization_columns <- matrix(c("rand0", "rand1", "rand2"), nrow = 1)
realization_index <- c("rand")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(t0, t1, lung) ~ loglinear(dose, rand, 0) + multiplicative()
res <- CoxRunMulti(formula, df,
  control = control,
  realization_columns = realization_columns,
  realization_index = realization_index
)

Automates creating a date difference column

Description

Date_Shift generates a new dataframe with a column containing time difference in a given unit

Usage

Date_Shift(df, dcol0, dcol1, col_name, units = "days")
Date_Shift(df, dcol0, dcol1, col_name, units = "days")

Arguments

df

a data.table containing the columns of interest

dcol0

list of starting month, day, and year

dcol1

list of ending month, day, and year

col_name

vector of new column names

units

time unit to use

Value

returns the updated dataframe

Examples

library(data.table)
m0 <- c(1, 1, 2, 2)
m1 <- c(2, 2, 3, 3)
d0 <- c(1, 2, 3, 4)
d1 <- c(6, 7, 8, 9)
y0 <- c(1990, 1991, 1997, 1998)
y1 <- c(2001, 2003, 2005, 2006)
df <- data.table(m0 = m0, m1 = m1, d0 = d0, d1 = d1, y0 = y0, y1 = y1)
df <- Date_Shift(df, c("m0", "d0", "y0"), c("m1", "d1", "y1"), "date_since")

library(data.table)
m0 <- c(1, 1, 2, 2)
m1 <- c(2, 2, 3, 3)
d0 <- c(1, 2, 3, 4)
d1 <- c(6, 7, 8, 9)
y0 <- c(1990, 1991, 1997, 1998)
y1 <- c(2001, 2003, 2005, 2006)
df <- data.table(m0 = m0, m1 = m1, d0 = d0, d1 = d1, y0 = y0, y1 = y1)
df <- Date_Shift(df, c("m0", "d0", "y0"), c("m1", "d1", "y1"), "date_since")

uses a table, list of categories, and list of event summaries to generate person-count tables

Description

Event_Count_Gen generates event-count tables

Usage

Event_Count_Gen(table, categ = list(), events = list())
Event_Count_Gen(table, categ = list(), events = list())

Arguments

table

dataframe with every category/event column needed

categ

list with category columns and methods, methods can be either strings or lists of boundaries

events

list of columns to summarize, supports counts and means and renaming the summary column

Value

returns a grouped table and a list of category boundaries used

Examples

library(data.table)
a <- c(0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7)
c <- c(0, 1, 0, 0, 0, 1, 0)
table <- data.table(
  a = a,
  b = b,
  c = c
)
categ <- list(
  a = "0/3/5]7",
  b = list(
    lower = c(-1, 3, 6),
    upper = c(3, 6, 10),
    name = c("low", "medium", "high")
  )
)
event <- list(
  c = "count AS cases",
  a = "mean", b = "mean"
)
e <- Event_Count_Gen(table, categ, event)
library(data.table)
a <- c(0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7)
c <- c(0, 1, 0, 0, 0, 1, 0)
table <- data.table(
  a = a,
  b = b,
  c = c
)
categ <- list(
  a = "0/3/5]7",
  b = list(
    lower = c(-1, 3, 6),
    upper = c(3, 6, 10),
    name = c("low", "medium", "high")
  )
)
event <- list(
  c = "count AS cases",
  a = "mean", b = "mean"
)
e <- Event_Count_Gen(table, categ, event)

uses a table, list of categories, list of summaries, list of events, and person-year information to generate person-time tables

Description

Event_Time_Gen generates event-time tables

Usage

Event_Time_Gen(
  table,
  pyr = list(),
  time_scale = list(),
  categ = list(),
  summaries = list(),
  events = NULL,
  fcount = FALSE,
  lcount = FALSE,
  studyid = "studyID",
  verbose = FALSE
)
Event_Time_Gen(
  table,
  pyr = list(),
  time_scale = list(),
  categ = list(),
  summaries = list(),
  events = NULL,
  fcount = FALSE,
  lcount = FALSE,
  studyid = "studyID",
  verbose = FALSE
)

Arguments

table

dataframe with every category/event column needed

pyr

list with entry and exit lists, containing day/month/year columns in the table

time_scale

list with the time scale information, either a calendar category or an age category

categ

list with category columns and methods, methods can be either strings or lists of boundaries

summaries

list of columns to summarize, supports counts, means, and weighted means by person-year and renaming the summary column

events

list of events or interests, checks if events are within each time interval

fcount

boolean if the ⁠first at risk⁠ count should be returned in the data. Returns the number of observations starting in each combination of categories.

lcount

boolean if the ⁠last at risk⁠ count should be returned in the data. Returns the number of observations ending in each combination of categories.

studyid

id used to determine distinct subjects used for first and last at risk intervals.

verbose

boolean if updates should be printed to the console.

Value

returns a grouped table and a list of category boundaries used

Examples

library(data.table)
a <- c(0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7)
c <- c(0, 1, 0, 0, 0, 1, 0)
d <- c(1, 2, 3, 4, 5, 6, 7)
e <- c(2, 3, 4, 5, 6, 7, 8)
f <- c(
  1900, 1900, 1900, 1900,
  1900, 1900, 1900
)
g <- c(1, 2, 3, 4, 5, 6, 7)
h <- c(2, 3, 4, 5, 6, 7, 8)
i <- c(
  1901, 1902, 1903, 1904,
  1905, 1906, 1907
)
table <- data.table(
  a = a, b = b, c = c,
  d = d, e = e, f = f,
  g = g, h = h, i = i
)
categ <- list(
  a = "-1/3/5]7"
)
calendar_categ <- list(
  type = "calendar",
  day = c(1, 1, 1),
  month = c(1, 1, 1),
  year = c(1901, 1904, 1908)
)
time_scale <- list(time = calendar_categ)
summary <- list(
  c = "count AS cases"
)
events <- list("c")
pyr <- list(
  entry = list(year = "f"),
  exit = list(year = "i"),
  unit = "years"
)
e <- Event_Time_Gen(table, pyr, time_scale, categ, summary, events)
library(data.table)
a <- c(0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7)
c <- c(0, 1, 0, 0, 0, 1, 0)
d <- c(1, 2, 3, 4, 5, 6, 7)
e <- c(2, 3, 4, 5, 6, 7, 8)
f <- c(
  1900, 1900, 1900, 1900,
  1900, 1900, 1900
)
g <- c(1, 2, 3, 4, 5, 6, 7)
h <- c(2, 3, 4, 5, 6, 7, 8)
i <- c(
  1901, 1902, 1903, 1904,
  1905, 1906, 1907
)
table <- data.table(
  a = a, b = b, c = c,
  d = d, e = e, f = f,
  g = g, h = h, i = i
)
categ <- list(
  a = "-1/3/5]7"
)
calendar_categ <- list(
  type = "calendar",
  day = c(1, 1, 1),
  month = c(1, 1, 1),
  year = c(1901, 1904, 1908)
)
time_scale <- list(time = calendar_categ)
summary <- list(
  c = "count AS cases"
)
events <- list("c")
pyr <- list(
  entry = list(year = "f"),
  exit = list(year = "i"),
  unit = "years"
)
e <- Event_Time_Gen(table, pyr, time_scale, categ, summary, events)

Generic background/excess event calculation function

Description

EventAssignment Generic background/excess event calculation function for standard poisson regressions and likelihood based confidence interval results.

Usage

EventAssignment(x, df, ...)
EventAssignment(x, df, ...)

Arguments

x

result object from a regression, class poisres

df

a data.table containing the columns of interest

...

extended for other necessary parameters

Value

returns matrices of events assigned to background and excess

Predicts how many events are due to baseline vs excess for a completed poisson model

Description

EventAssignment.poisres uses user provided data, person-year/event columns, vectors specifying the model, and options to calculate background and excess events for a solved Poisson regression

Usage

## S3 method for class 'poisres'
EventAssignment(
  x,
  df,
  assign_control = list(),
  control = list(),
  a_n = NULL,
  ...
)
## S3 method for class 'poisres'
EventAssignment(
  x,
  df,
  assign_control = list(),
  control = list(),
  a_n = NULL,
  ...
)

Arguments

x

result object from a regression, class poisres

df

a data.table containing the columns of interest

assign_control

control list for bounds calculated

control

list of parameters controlling the convergence, see the control options vignette for details

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the assign_control list parameter

Value

returns matrices of events assigned to background and excess

Predicts how many events are due to baseline vs excess for a completed poisson likelihood boundary regression

Description

EventAssignment.poisresbound uses user provided data, person-year/event columns, vectors specifying the model, and options to calculate background and excess events for a solved Poisson regression

Usage

## S3 method for class 'poisresbound'
EventAssignment(
  x,
  df,
  assign_control = list(),
  control = list(),
  a_n = NULL,
  ...
)
## S3 method for class 'poisresbound'
EventAssignment(
  x,
  df,
  assign_control = list(),
  control = list(),
  a_n = NULL,
  ...
)

Arguments

x

result object from a regression, class poisres

df

a data.table containing the columns of interest

assign_control

control list for bounds calculated

control

list of parameters controlling the convergence, see the control options vignette for details

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the assign_control list parameter

Value

returns matrices of events assigned to background and excess

Splits a parameter into factors

Description

factorize uses user provided list of columns to define new parameter for each unique value and update the data.table. Not for interaction terms

Usage

factorize(df, col_list, verbose = 0)
factorize(df, col_list, verbose = 0)

Arguments

df

a data.table containing the columns of interest

col_list

an array of column names that should have factor terms defined

verbose

integer valued 0-4 controlling what information is printed to the terminal. Each level includes the lower levels. 0: silent, 1: errors printed, 2: warnings printed, 3: notes printed, 4: debug information printed. Errors are situations that stop the regression, warnings are situations that assume default values that the user might not have intended, notes provide information on regression progress, and debug prints out C++ progress and intermediate results. The default level is 2 and True/False is converted to 3/0.

Value

returns a list with two named fields. df for the updated dataframe, and cols for the new column names

Examples

library(data.table)
a <- c(0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7)
c <- c(0, 1, 2, 1, 0, 1, 0)
df <- data.table(a = a, b = b, c = c)
col_list <- c("c")
val <- factorize(df, col_list)
df <- val$df
new_col <- val$cols
library(data.table)
a <- c(0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7)
c <- c(0, 1, 2, 1, 0, 1, 0)
df <- data.table(a = a, b = b, c = c)
col_list <- c("c")
val <- factorize(df, col_list)
df <- val$df
new_col <- val$cols

Applies time dependence to parameters

Description

gen_time_dep generates a new dataframe with time dependent covariates by applying a grid in time

Usage

gen_time_dep(
  df,
  time1,
  time2,
  event0,
  iscox,
  dt,
  new_names,
  dep_cols,
  func_form,
  fname,
  tform,
  nthreads = as.numeric(detectCores())
)
gen_time_dep(
  df,
  time1,
  time2,
  event0,
  iscox,
  dt,
  new_names,
  dep_cols,
  func_form,
  fname,
  tform,
  nthreads = as.numeric(detectCores())
)

Arguments

df

a data.table containing the columns of interest

time1

column used for time period starts

time2

column used for time period end

event0

column used for event status

iscox

boolean if rows not at event times should not be kept, rows are removed if true. a Cox proportional hazards model does not use rows with intervals not containing event times

dt

spacing in time for new rows

new_names

list of new names to use instead of default, default used if entry is ”"

dep_cols

columns that are not needed in the new dataframe

func_form

vector of functions to apply to each time-dependent covariate. Of the form func(df, time) returning a vector of the new column value

fname

filename used for new dataframe

tform

list of string function identifiers, used for linear/step

nthreads

number of threads to use, do not use more threads than available on your machine

Value

returns the updated dataframe

Examples

library(data.table)
# Adapted from the tests
a <- c(20, 20, 5, 10, 15)
b <- c(1, 2, 1, 1, 2)
c <- c(0, 0, 1, 1, 1)
df <- data.table(a = a, b = b, c = c)
time1 <- "%trunc%"
time2 <- "a"
event <- "c"
grt_f <- function(df, time_col) {
  return((df[, "b"] * df[, get(time_col)])[[1]])
}
func_form <- "lin"
df_new <- gen_time_dep(
  df, time1, time2, event, TRUE, 0.01, c("grt"), NULL,
  c(grt_f), paste0(tempfile(), "test_new.csv"), func_form, 1
)
library(data.table)
# Adapted from the tests
a <- c(20, 20, 5, 10, 15)
b <- c(1, 2, 1, 1, 2)
c <- c(0, 0, 1, 1, 1)
df <- data.table(a = a, b = b, c = c)
time1 <- "%trunc%"
time2 <- "a"
event <- "c"
grt_f <- function(df, time_col) {
  return((df[, "b"] * df[, get(time_col)])[[1]])
}
func_form <- "lin"
df_new <- gen_time_dep(
  df, time1, time2, event, TRUE, 0.01, c("grt"), NULL,
  c(grt_f), paste0(tempfile(), "test_new.csv"), func_form, 1
)

Interprets a Colossus formula and makes necessary changes to data

Description

get_form uses a formula and data.table, to fully describe the model for a Colossus regression function.

Usage

get_form(formula, df, nthreads = as.numeric(detectCores())/2)
get_form(formula, df, nthreads = as.numeric(detectCores())/2)

Arguments

formula

a formula object, written in Colossus notation. See the unified equation expression vignette vignette for details.

df

a data.table containing the columns of interest

nthreads

number of threads to use, do not use more threads than available on your machine

Value

returns a class fully describing the model and the updated data

Examples

library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
model <- get_form(formula, df, 1)
library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
model <- get_form(formula, df, 1)

Interprets a Poisson joint formula and makes necessary changes to data

Description

get_form_joint uses two event formula, a shared formula, and data.table, to fully describe the model for a joint Poisson model.

Usage

get_form_joint(formula_list, df, nthreads = as.numeric(detectCores())/2)
get_form_joint(formula_list, df, nthreads = as.numeric(detectCores())/2)

Arguments

formula_list

a list of formula objects, each written in Colossus notation. See the unified equation expression vignette vignette for details. Each formula should include the elements specific to the specified event column. The list can include an entry named "shared" to denote shared terms. The person-year and strata columns should be the same.

df

a data.table containing the columns of interest

nthreads

number of threads to use, do not use more threads than available on your machine

Value

returns a class fully describing the model and the updated data

Automates creating data for a joint competing risks analysis

Description

Joint_Multiple_Events generates input for a regression with multiple non-independent events and models

Usage

Joint_Multiple_Events(
  df,
  events,
  name_list,
  term_n_list = list(),
  tform_list = list(),
  keep_constant_list = list(),
  a_n_list = list()
)
Joint_Multiple_Events(
  df,
  events,
  name_list,
  term_n_list = list(),
  tform_list = list(),
  keep_constant_list = list(),
  a_n_list = list()
)

Arguments

df

a data.table containing the columns of interest

events

vector of event column names

name_list

list of vectors for columns for event specific or shared model elements, required

term_n_list

list of vectors for term numbers for event specific or shared model elements, defaults to term 0

tform_list

list of vectors for subterm types for event specific or shared model elements, defaults to loglinear

keep_constant_list

list of vectors for constant elements for event specific or shared model elements, defaults to free (0)

a_n_list

list of vectors for parameter values for event specific or shared model elements, defaults to term 0

Value

returns the updated dataframe and model inputs

Examples

library(data.table)
a <- c(0, 0, 0, 1, 1, 1)
b <- c(1, 1, 1, 2, 2, 2)
c <- c(0, 1, 2, 2, 1, 0)
d <- c(1, 1, 0, 0, 1, 1)
e <- c(0, 1, 1, 1, 0, 0)
df <- data.table(t0 = a, t1 = b, e0 = c, e1 = d, fac = e)
time1 <- "t0"
time2 <- "t1"
df$pyr <- df$t1 - df$t0
pyr <- "pyr"
events <- c("e0", "e1")
names_e0 <- c("fac")
names_e1 <- c("fac")
names_shared <- c("t0", "t0")
term_n_e0 <- 0
term_n_e1 <- 0
term_n_shared <- c(0, 0)
tform_e0 <- "loglin"
tform_e1 <- "loglin"
tform_shared <- c("quad_slope", "loglin_top")
keep_constant_e0 <- 0
keep_constant_e1 <- 0
keep_constant_shared <- c(0, 0)
a_n_e0 <- c(-0.1)
a_n_e1 <- c(0.1)
a_n_shared <- c(0.001, -0.02)
name_list <- list(shared = names_shared, e0 = names_e0, e1 = names_e1)
term_n_list <- list(shared = term_n_shared, e0 = term_n_e0, e1 = term_n_e1)
tform_list <- list(shared = tform_shared, e0 = tform_e0, e1 = tform_e1)
keep_constant_list <- list(
  shared = keep_constant_shared,
  e0 = keep_constant_e0, e1 = keep_constant_e1
)
a_n_list <- list(shared = a_n_shared, e0 = a_n_e0, e1 = a_n_e1)
val <- Joint_Multiple_Events(
  df, events, name_list, term_n_list,
  tform_list, keep_constant_list, a_n_list
)

library(data.table)
a <- c(0, 0, 0, 1, 1, 1)
b <- c(1, 1, 1, 2, 2, 2)
c <- c(0, 1, 2, 2, 1, 0)
d <- c(1, 1, 0, 0, 1, 1)
e <- c(0, 1, 1, 1, 0, 0)
df <- data.table(t0 = a, t1 = b, e0 = c, e1 = d, fac = e)
time1 <- "t0"
time2 <- "t1"
df$pyr <- df$t1 - df$t0
pyr <- "pyr"
events <- c("e0", "e1")
names_e0 <- c("fac")
names_e1 <- c("fac")
names_shared <- c("t0", "t0")
term_n_e0 <- 0
term_n_e1 <- 0
term_n_shared <- c(0, 0)
tform_e0 <- "loglin"
tform_e1 <- "loglin"
tform_shared <- c("quad_slope", "loglin_top")
keep_constant_e0 <- 0
keep_constant_e1 <- 0
keep_constant_shared <- c(0, 0)
a_n_e0 <- c(-0.1)
a_n_e1 <- c(0.1)
a_n_shared <- c(0.001, -0.02)
name_list <- list(shared = names_shared, e0 = names_e0, e1 = names_e1)
term_n_list <- list(shared = term_n_shared, e0 = term_n_e0, e1 = term_n_e1)
tform_list <- list(shared = tform_shared, e0 = tform_e0, e1 = tform_e1)
keep_constant_list <- list(
  shared = keep_constant_shared,
  e0 = keep_constant_e0, e1 = keep_constant_e1
)
a_n_list <- list(shared = a_n_shared, e0 = a_n_e0, e1 = a_n_e1)
val <- Joint_Multiple_Events(
  df, events, name_list, term_n_list,
  tform_list, keep_constant_list, a_n_list
)

Defines the likelihood ratio test

Description

Likelihood_Ratio_Test uses two models and calculates the ratio

Usage

Likelihood_Ratio_Test(alternative_model, null_model)
Likelihood_Ratio_Test(alternative_model, null_model)

Arguments

alternative_model

the new model of interest in list form, output from a Poisson regression

null_model

a model to compare against, in list form

Value

returns the score statistic

Examples

library(data.table)
# In an actual example, one would run two seperate RunCoxRegression regressions,
#    assigning the results to e0 and e1
a <- c(0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7, 1, 2, 3, 4, 5, 6, 7)
c <- c(1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0)
d <- c(3, 4, 5, 6, 7, 8, 9, 1, 2, 1, 1, 2, 1, 2)
e <- c(1, 2, 0, 0, 1, 2, 0, 0, 1, 2, 0, 0, 1, 2)
df <- data.table(a = a, b = b, c = c, d = d, e = e)
keep_constant <- 0
a_n <- c(-0.1, 0.1, 0.1, 0.2)
control <- list(ncores = 1, maxiter = 10, verbose = 0)
model <- Cox(a, b, c) ~ plinear(d * d, 0) + loglinear(factor(e))
alternative_model <- CoxRun(model, df,
  control = control,
  a_n = a_n, keep_constant = c(0, 1, 0)
)
null_model <- CoxRun(Cox(a, b, c) ~ null(), df, control = control)
score <- Likelihood_Ratio_Test(alternative_model, null_model)
library(data.table)
# In an actual example, one would run two seperate RunCoxRegression regressions,
#    assigning the results to e0 and e1
a <- c(0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6)
b <- c(1, 2, 3, 4, 5, 6, 7, 1, 2, 3, 4, 5, 6, 7)
c <- c(1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0)
d <- c(3, 4, 5, 6, 7, 8, 9, 1, 2, 1, 1, 2, 1, 2)
e <- c(1, 2, 0, 0, 1, 2, 0, 0, 1, 2, 0, 0, 1, 2)
df <- data.table(a = a, b = b, c = c, d = d, e = e)
keep_constant <- 0
a_n <- c(-0.1, 0.1, 0.1, 0.2)
control <- list(ncores = 1, maxiter = 10, verbose = 0)
model <- Cox(a, b, c) ~ plinear(d * d, 0) + loglinear(factor(e))
alternative_model <- CoxRun(model, df,
  control = control,
  a_n = a_n, keep_constant = c(0, 1, 0)
)
null_model <- CoxRun(Cox(a, b, c) ~ null(), df, control = control)
score <- Likelihood_Ratio_Test(alternative_model, null_model)

Generic likelihood boundary calculation function

Description

LikelihoodBound Generic likelihood boundary calculation function. Used by the coxres, poisres, and logitres objects to calculate likelihood based confidence intervals.

Usage

LikelihoodBound(x, df, curve_control = list(), control = list(), ...)
LikelihoodBound(x, df, curve_control = list(), control = list(), ...)

Arguments

x

result object from a regression, class coxres or poisres

df

a data.table containing the columns of interest

curve_control

a list of control options for the likelihood boundary regression. See the control options vignette for details.

control

list of parameters controlling the convergence, see the control options vignette for details

...

extended for other necessary parameters

Value

returns a summary of the likelihood boundary

Calculates the likelihood boundary for a completed cox model

Description

LikelihoodBound.coxres solves the confidence interval for a cox model, starting at the optimum point and iteratively optimizing end-points of intervals.

Usage

## S3 method for class 'coxres'
LikelihoodBound(x, df, curve_control = list(), control = list(), ...)
## S3 method for class 'coxres'
LikelihoodBound(x, df, curve_control = list(), control = list(), ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

curve_control

a list of control options for the likelihood boundary regression. See the control options vignette for details.

control

list of parameters controlling the convergence, see the control options vignette for details

...

can include the named entries for the curve_control list parameter

Value

returns a summary of the likelihood boundary

Calculates the likelihood boundary for a completed Logistic model

Description

LikelihoodBound.logitres solves the confidence interval for a Poisson model, starting at the optimum point and iteratively optimizing end-points of intervals.

Usage

## S3 method for class 'logitres'
LikelihoodBound(x, df, curve_control = list(), control = list(), ...)
## S3 method for class 'logitres'
LikelihoodBound(x, df, curve_control = list(), control = list(), ...)

Arguments

x

result object from a regression, class poisres

df

a data.table containing the columns of interest

curve_control

a list of control options for the likelihood boundary regression. See the control options vignette for details.

control

list of parameters controlling the convergence, see the control options vignette for details

...

can include the named entries for the curve_control list parameter

Value

returns a summary of the likelihood boundary

Calculates the likelihood boundary for a completed Poisson model

Description

LikelihoodBound.poisres solves the confidence interval for a Poisson model, starting at the optimum point and iteratively optimizing end-points of intervals.

Usage

## S3 method for class 'poisres'
LikelihoodBound(x, df, curve_control = list(), control = list(), ...)
## S3 method for class 'poisres'
LikelihoodBound(x, df, curve_control = list(), control = list(), ...)

Arguments

x

result object from a regression, class poisres

df

a data.table containing the columns of interest

curve_control

a list of control options for the likelihood boundary regression. See the control options vignette for details.

control

list of parameters controlling the convergence, see the control options vignette for details

...

can include the named entries for the curve_control list parameter

Value

returns a summary of the likelihood boundary

Calculates Full Parameter list for Special Dose Formula

Description

Linked_Dose_Formula Calculates all parameters for linear-quadratic and linear-exponential linked formulas

Usage

Linked_Dose_Formula(tforms, paras)
Linked_Dose_Formula(tforms, paras)

Arguments

tforms

list of formula types

paras

list of formula parameters

Value

returns list of full parameters

Examples

library(data.table)
tforms <- list(cov_0 = "quad", cov_1 = "exp")
paras <- list(cov_0 = c(1, 3.45), cov_1 = c(1.2, 4.5, 0.1))
full_paras <- Linked_Dose_Formula(tforms, paras)
library(data.table)
tforms <- list(cov_0 = "quad", cov_1 = "exp")
paras <- list(cov_0 = c(1, 3.45), cov_1 = c(1.2, 4.5, 0.1))
full_paras <- Linked_Dose_Formula(tforms, paras)

Calculates The Additional Parameter For a linear-exponential formula with known maximum

Description

Linked_Lin_Exp_Para Calculates what the additional parameter would be for a desired maximum

Usage

Linked_Lin_Exp_Para(y, a0, a1_goal)
Linked_Lin_Exp_Para(y, a0, a1_goal)

Arguments

y

point formula switch

a0

linear slope

a1_goal

exponential maximum desired

Value

returns parameter used by Colossus

Examples

library(data.table)
y <- 7.6
a0 <- 1.2
a1_goal <- 15
full_paras <- Linked_Lin_Exp_Para(y, a0, a1_goal)
library(data.table)
y <- 7.6
a0 <- 1.2
a1_goal <- 15
full_paras <- Linked_Lin_Exp_Para(y, a0, a1_goal)

Fully runs a logistic regression model, returning the model and results

Description

LogisticRun uses a formula, data.table, and list of controls to prepare and run a Colossus logistic regression function

Usage

LogisticRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  link = "odds",
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)
LogisticRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  link = "odds",
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

link

Used in logistic regression, the linking function relating the input model and event probability. Current options are "odds", "ident", and "loglink" for the odds ratio, identity, and complimentary loglink options.

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

norm

methods used to normalize the covariates. Default is 'null' for no normalization. Other options include 'max' to normalize by the absolute maximum and 'mean' to normalize by the mean

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- logit(Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- LogisticRun(formula, df, a_n = c(1.1, -0.1, 0.2, 0.5), control = control)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- logit(Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- LogisticRun(formula, df, a_n = c(1.1, -0.1, 0.2, 0.5), control = control)

Fully runs a logistic regression model with multiple column realizations, returning the model and results

Description

LogisticRunMulti uses a formula, data.table, and list of controls to prepare and run a Colossus poisson regression function

Usage

LogisticRunMulti(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = matrix(c("temp00", "temp01", "temp10", "temp11"), nrow = 2),
  realization_index = c("temp0", "temp1"),
  control = list(),
  gradient_control = list(),
  link = "odds",
  single = FALSE,
  observed_info = FALSE,
  fma = TRUE,
  mcml = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)
LogisticRunMulti(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = matrix(c("temp00", "temp01", "temp10", "temp11"), nrow = 2),
  realization_index = c("temp0", "temp1"),
  control = list(),
  gradient_control = list(),
  link = "odds",
  single = FALSE,
  observed_info = FALSE,
  fma = TRUE,
  mcml = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

realization_columns

used for multi-realization regressions. Matrix of column names with rows for each column with realizations, columns for each realization

realization_index

used for multi-realization regressions. Vector of column names, one for each column with realizations. Each name should be used in the "names" variable in the equation definition

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

link

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

fma

a boolean to denote that the Frequentist Model Averaging method should be used

mcml

a boolean to denote that the Monte Carlo Maximum Likelihood method should be used

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- floor(runif(nrow(df), min = 0, max = 5))
df$rand0 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand1 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand2 <- floor(runif(nrow(df), min = 0, max = 5))
realization_columns <- matrix(c("rand0", "rand1", "rand2"), nrow = 1)
realization_index <- c("rand")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- logit(lung) ~ loglinear(CONST, dose, rand, 0) + multiplicative()
res <- LogisticRunMulti(formula, df,
  control = control,
  realization_columns = realization_columns,
  realization_index = realization_index
)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- floor(runif(nrow(df), min = 0, max = 5))
df$rand0 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand1 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand2 <- floor(runif(nrow(df), min = 0, max = 5))
realization_columns <- matrix(c("rand0", "rand1", "rand2"), nrow = 1)
realization_index <- c("rand")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- logit(lung) ~ loglinear(CONST, dose, rand, 0) + multiplicative()
res <- LogisticRunMulti(formula, df,
  control = control,
  realization_columns = realization_columns,
  realization_index = realization_index
)

Fully runs a logistic regression model with multiple event realizations, returning the model and results

Description

LogisticRunMultiOut uses a formula, data.table, and list of controls to prepare and run a Colossus logistic regression function

Usage

LogisticRunMultiOut(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = c("event0", "event1"),
  control = list(),
  gradient_control = list(),
  link = "odds",
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)
LogisticRunMultiOut(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = c("event0", "event1"),
  control = list(),
  gradient_control = list(),
  link = "odds",
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

realization_columns

used for multi-realization regressions. Matrix of column names with rows for each column with realizations, columns for each realization

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

link

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  event = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$event0 <- rbinom(nrow(df), size = 1, prob = 0.3)
df$event1 <- rbinom(nrow(df), size = 1, prob = 0.3)
df$event2 <- rbinom(nrow(df), size = 1, prob = 0.3)
realization_columns <- c("event0", "event1", "event2")
control <- list(
  ncores = 1, lr = 0.75, maxiter = 10,
  verbose = 0
)
formula <- logit(event) ~ loglinear(dose, 0) + multiplicative()
res <- LogisticRunMultiOut(formula, df,
  realization_columns = realization_columns,
  control = control
)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  event = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$event0 <- rbinom(nrow(df), size = 1, prob = 0.3)
df$event1 <- rbinom(nrow(df), size = 1, prob = 0.3)
df$event2 <- rbinom(nrow(df), size = 1, prob = 0.3)
realization_columns <- c("event0", "event1", "event2")
control <- list(
  ncores = 1, lr = 0.75, maxiter = 10,
  verbose = 0
)
formula <- logit(event) ~ loglinear(dose, 0) + multiplicative()
res <- LogisticRunMultiOut(formula, df,
  realization_columns = realization_columns,
  control = control
)

Checks the OMP flag

Description

OMP_Check Called directly from R, checks the omp flag and returns true if omp is enabled

Usage

OMP_Check()
OMP_Check()

Value

boolean: True for OMP allowed

Performs Cox Proportional Hazard model plots

Description

plot.coxres uses user provided data, time/event columns, vectors specifying the model, and options to choose and save plots. The 'plot_options' variable can be used to select the types of plots run.

Usage

## S3 method for class 'coxres'
plot(x, df, plot_options, a_n = NULL, ...)
## S3 method for class 'coxres'
plot(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Examples

library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  control = control,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4))
)
plot_options <- list(
  type = c("surv", paste0(
    tempfile(),
    "run"
  )), studyid = "UserID",
  verbose = FALSE
)
res_plot <- plot(res, df, plot_options)
library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  control = control,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4))
)
plot_options <- list(
  type = c("surv", paste0(
    tempfile(),
    "run"
  )), studyid = "UserID",
  verbose = FALSE
)
res_plot <- plot(res, df, plot_options)

Generic Martingale Residual Plotting function

Description

plotMartingale Generic martingale residual plotting, used by the coxres object to plot martingale residuals. Currently only returns data that could be plotted.

Usage

plotMartingale(x, df, plot_options, a_n = NULL, ...)
plotMartingale(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Performs Cox Proportional Hazard model martingale residual plots

Description

plotMartingale.coxres uses user provided data, time/event columns, vectors specifying the model, and options to choose and save plots

Usage

## S3 method for class 'coxres'
plotMartingale(x, df, plot_options, a_n = NULL, ...)
## S3 method for class 'coxres'
plotMartingale(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Generic Risk Plotting function

Description

plotRisk Generic risk plotting function. Used by the coxres object to plot the hazard ratios at different levels of individual covariates. Currently only returns data that could be plotted.

Usage

plotRisk(x, df, plot_options, a_n = NULL, ...)
plotRisk(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Performs Cox Proportional Hazard model hazard ratio plots

Description

plotRisk.coxres uses user provided data, time/event columns, vectors specifying the model, and options to choose and save plots

Usage

## S3 method for class 'coxres'
plotRisk(x, df, plot_options, a_n = NULL, ...)
## S3 method for class 'coxres'
plotRisk(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Generic Schoenfeld Residual Plotting function

Description

plotSchoenfeld Generic Schoenfeld residual plotting function, used by the coxres object to calculate Schoenfeld residuals. Currently only returns data that could be plotted.

Usage

plotSchoenfeld(x, df, plot_options, a_n = NULL, ...)
plotSchoenfeld(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Performs Cox Proportional Hazard model schoenfeld residual plots

Description

plotSchoenfeld.coxres uses user provided data, time/event columns, vectors specifying the model, and options to choose and save plots

Usage

## S3 method for class 'coxres'
plotSchoenfeld(x, df, plot_options, a_n = NULL, ...)
## S3 method for class 'coxres'
plotSchoenfeld(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Generic Survival Plotting function

Description

plotSurvival Generic survival plotting, used by the coxres object. Calculates the baseline hazard, cumulative hazard, and survival curve for the list of event times. Currently only returns data that could be plotted.

Usage

plotSurvival(x, df, plot_options, a_n = NULL, ...)
plotSurvival(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Performs Cox Proportional Hazard model survival plots

Description

plotSurvival.coxres uses user provided data, time/event columns, vectors specifying the model, and options to choose and save plots

Usage

## S3 method for class 'coxres'
plotSurvival(x, df, plot_options, a_n = NULL, ...)
## S3 method for class 'coxres'
plotSurvival(x, df, plot_options, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

plot_options

list of parameters controlling the plot options, see the plotting vignette for different options

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

can include the named entries for the plot_options parameter

Value

returns the data used for plots

Examples

library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  control = control,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4))
)
plot_options <- list(
  fname = paste0(
    tempfile(),
    "run"
  ), studyid = "UserID",
  verbose = FALSE
)
res_plot <- plotSurvival(res, df, plot_options)
library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  control = control,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4))
)
plot_options <- list(
  fname = paste0(
    tempfile(),
    "run"
  ), studyid = "UserID",
  verbose = FALSE
)
res_plot <- plotSurvival(res, df, plot_options)

Fully runs a poisson regression model, returning the model and results

Description

PoisRun uses a formula, data.table, and list of controls to prepare and run a Colossus poisson regression function

Usage

PoisRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)
PoisRun(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

norm

methods used to normalize the covariates. Default is 'null' for no normalization. Other options include 'max' to normalize by the absolute maximum and 'mean' to normalize by the mean

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Pois(Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- PoisRun(formula, df, a_n = c(1.1, -0.1, 0.2, 0.5), control = control)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Pois(Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- PoisRun(formula, df, a_n = c(1.1, -0.1, 0.2, 0.5), control = control)

Fully runs a joint poisson regression model, returning the model and results

Description

PoisRunJoint uses a list of formula, data.table, and list of controls to prepare and run a Colossus poisson regression function on a joint dataset

Usage

PoisRunJoint(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)
PoisRunJoint(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  norm = "null",
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

norm

methods used to normalize the covariates. Default is 'null' for no normalization. Other options include 'max' to normalize by the absolute maximum and 'mean' to normalize by the mean

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  Flu_Status = c(0, 1, 0, 0, 1, 0, 1),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula_list <- list(Pois(Ending_Age, Cancer_Status) ~ plinear(d, 0),
  Pois(Ending_Age, Flu_Status) ~ loglinear(d, 0),
  shared = Pois(Ending_Age) ~ loglinear(a, b, c, 0)
)
res <- PoisRunJoint(formula_list, df, control = control)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  Flu_Status = c(0, 1, 0, 0, 1, 0, 1),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula_list <- list(Pois(Ending_Age, Cancer_Status) ~ plinear(d, 0),
  Pois(Ending_Age, Flu_Status) ~ loglinear(d, 0),
  shared = Pois(Ending_Age) ~ loglinear(a, b, c, 0)
)
res <- PoisRunJoint(formula_list, df, control = control)

Fully runs a poisson regression model with multiple column realizations, returning the model and results

Description

PoisRunMulti uses a formula, data.table, and list of controls to prepare and run a Colossus poisson regression function

Usage

PoisRunMulti(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = matrix(c("temp00", "temp01", "temp10", "temp11"), nrow = 2),
  realization_index = c("temp0", "temp1"),
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  fma = TRUE,
  mcml = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)
PoisRunMulti(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = matrix(c("temp00", "temp01", "temp10", "temp11"), nrow = 2),
  realization_index = c("temp0", "temp1"),
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  fma = TRUE,
  mcml = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

realization_columns

used for multi-realization regressions. Matrix of column names with rows for each column with realizations, columns for each realization

realization_index

used for multi-realization regressions. Vector of column names, one for each column with realizations. Each name should be used in the "names" variable in the equation definition

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

fma

a boolean to denote that the Frequentist Model Averaging method should be used

mcml

a boolean to denote that the Monte Carlo Maximum Likelihood method should be used

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- floor(runif(nrow(df), min = 0, max = 5))
df$rand0 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand1 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand2 <- floor(runif(nrow(df), min = 0, max = 5))
realization_columns <- matrix(c("rand0", "rand1", "rand2"), nrow = 1)
realization_index <- c("rand")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Pois(t1, lung) ~ loglinear(CONST, dose, rand, 0) + multiplicative()
res <- PoisRunMulti(formula, df,
  control = control,
  realization_columns = realization_columns,
  realization_index = realization_index
)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- floor(runif(nrow(df), min = 0, max = 5))
df$rand0 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand1 <- floor(runif(nrow(df), min = 0, max = 5))
df$rand2 <- floor(runif(nrow(df), min = 0, max = 5))
realization_columns <- matrix(c("rand0", "rand1", "rand2"), nrow = 1)
realization_index <- c("rand")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Pois(t1, lung) ~ loglinear(CONST, dose, rand, 0) + multiplicative()
res <- PoisRunMulti(formula, df,
  control = control,
  realization_columns = realization_columns,
  realization_index = realization_index
)

Fully runs a poisson regression model with multiple event realizations, returning the model and results

Description

PoisRunMultiOut uses a formula, data.table, and list of controls to prepare and run a Colossus poisson regression function

Usage

PoisRunMultiOut(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = c("event0", "event1"),
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)
PoisRunMultiOut(
  model,
  df,
  a_n = list(0),
  keep_constant = 0,
  realization_columns = c("event0", "event1"),
  control = list(),
  gradient_control = list(),
  single = FALSE,
  observed_info = FALSE,
  cons_mat = as.matrix(0),
  cons_vec = 0,
  ...
)

Arguments

model

either a formula written for the get_form function, or the model result from the get_form function.

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

keep_constant

binary values to denote which parameters to change

realization_columns

used for multi-realization regressions. Matrix of column names with rows for each column with realizations, columns for each realization

control

list of parameters controlling the convergence, see the control options vignette for details

gradient_control

a list of control options for the gradient descent algorithm. If any value is given, a gradient descent algorithm is used instead of Newton-Raphson. See the control options vignette for details

single

a boolean to denote that only the log-likelihood should be calculated and returned, no derivatives or iterations

observed_info

a boolean to denote that the observed information matrix should be used to calculate the standard error for parameters, not the expected information matrix

cons_mat

Matrix containing coefficients for a system of linear constraints, formatted as matrix

cons_vec

Vector containing constants for a system of linear constraints, formatted as vector

...

can include the named entries for the control list parameter

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
df$lung0 <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
df$lung1 <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
df$lung2 <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
realization_columns <- c("lung", "lung1", "lung2")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Pois(t1, lung) ~ loglinear(CONST, dose, rand, 0) + multiplicative()
res <- PoisRunMultiOut(formula, df,
  control = control,
  realization_columns = realization_columns
)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  t0 = c(18, 20, 18, 19, 21, 20, 18),
  t1 = c(30, 45, 57, 47, 36, 60, 55),
  lung = c(0, 0, 1, 0, 1, 0, 0),
  dose = c(0, 1, 1, 0, 1, 0, 1)
)
set.seed(3742)
df$rand <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
df$lung0 <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
df$lung1 <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
df$lung2 <- c(1, floor(runif(nrow(df) - 1, min = 0, max = 1)))
realization_columns <- c("lung", "lung1", "lung2")
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Pois(t1, lung) ~ loglinear(CONST, dose, rand, 0) + multiplicative()
res <- PoisRunMultiOut(formula, df,
  control = control,
  realization_columns = realization_columns
)

Generic relative risk calculation function

Description

RelativeRisk Generic relative risk calculation function, currently only used by the coxres object, to calculate relative risks for either the last parameter or a custom point.

Usage

RelativeRisk(x, df, ...)
RelativeRisk(x, df, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

...

extended for other necessary parameters

Value

returns matrices of the relatives risks

Calculates hazard ratios for a reference vector and model

Description

RelativeRisk.coxmodel uses a cox model object and data, to evaluate relative risk in the data using the risk model from the result

Usage

## S3 method for class 'coxmodel'
RelativeRisk(x, df, a_n = NULL, ...)
## S3 method for class 'coxmodel'
RelativeRisk(x, df, a_n = NULL, ...)

Arguments

x

cox model object, which could be fed to a regression

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

extended to match any future parameters needed

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
model <- get_form(formula, df)$model
res_risk <- RelativeRisk(model, df, a_n = c(1.1, -0.1, 0.2, 0.5))
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
model <- get_form(formula, df)$model
res_risk <- RelativeRisk(model, df, a_n = c(1.1, -0.1, 0.2, 0.5))

Calculates hazard ratios for a reference vector

Description

coxres.RelativeRisk uses a cox result object and data, to evaluate relative risk in the data using the risk model from the result

Usage

## S3 method for class 'coxres'
RelativeRisk(x, df, a_n = NULL, ...)
## S3 method for class 'coxres'
RelativeRisk(x, df, a_n = NULL, ...)

Arguments

x

result object from a regression, class coxres

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

extended to match any future parameters needed

Value

returns a class fully describing the model and the regression results

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4)),
  control = control
)
res_risk <- RelativeRisk(res, df)
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res <- CoxRun(formula, df,
  a_n = list(c(1.1, -0.1, 0.2, 0.5), c(1.6, -0.12, 0.3, 0.4)),
  control = control
)
res_risk <- RelativeRisk(res, df)

Calculates hazard ratios for a reference vector and model

Description

RelativeRisk.formula uses a formula and data, to evaluate relative risk in the data using the risk model from the result

Usage

## S3 method for class 'formula'
RelativeRisk(x, df, a_n = NULL, ...)
## S3 method for class 'formula'
RelativeRisk(x, df, a_n = NULL, ...)

Arguments

x

formula object, which could be fed to a regression

df

a data.table containing the columns of interest

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

...

extended to match any future parameters needed

Value

returns matrices of the relatives risks

Examples

library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res_risk <- RelativeRisk(formula, df, a_n = c(1.1, -0.1, 0.2, 0.5))
library(data.table)
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, 57, 47, 36, 60, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0),
  a = c(0, 1, 1, 0, 1, 0, 1),
  b = c(1, 1.1, 2.1, 2, 0.1, 1, 0.2),
  c = c(10, 11, 10, 11, 12, 9, 11),
  d = c(0, 0, 0, 1, 1, 1, 1),
  e = c(0, 0, 1, 0, 0, 0, 1)
)
control <- list(
  ncores = 1, lr = 0.75, maxiters = c(1, 1),
  halfmax = 1
)
formula <- Cox(Starting_Age, Ending_Age, Cancer_Status) ~
  loglinear(a, b, c, 0) + plinear(d, 0) + multiplicative()
res_risk <- RelativeRisk(formula, df, a_n = c(1.1, -0.1, 0.2, 0.5))

Automatically assigns missing values in listed columns

Description

Replace_Missing checks each column and fills in NA values

Usage

Replace_Missing(df, name_list, msv, verbose = FALSE)
Replace_Missing(df, name_list, msv, verbose = FALSE)

Arguments

df

a data.table containing the columns of interest

name_list

vector of string column names to check

msv

value to replace na with, same used for every column used

verbose

Value

returns a filled datatable

Examples

library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, NA, 47, 36, NA, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0)
)
df <- Replace_Missing(df, c("Starting_Age", "Ending_Age"), 70)
library(data.table)
## basic example code reproduced from the starting-description vignette
df <- data.table(
  UserID = c(112, 114, 213, 214, 115, 116, 117),
  Starting_Age = c(18, 20, 18, 19, 21, 20, 18),
  Ending_Age = c(30, 45, NA, 47, 36, NA, 55),
  Cancer_Status = c(0, 0, 1, 0, 1, 0, 0)
)
df <- Replace_Missing(df, c("Starting_Age", "Ending_Age"), 70)

Generic Residual calculation function

Description

Residual Generic residual calculation function for regression results. Note that the Cox residuals are calculated using the various plotting based functions.

Usage

Residual(x, df, ...)
Residual(x, df, ...)

Arguments

x

result object from a regression, class coxres or poisres

df

a data.table containing the columns of interest

...

extended for other necessary parameters

Value

returns matrices of residuals

Calculates the Residuals for a logistic model

Description

Residual.logitmodel uses user provided data, person-year/event columns, vectors specifying the model, and options to calculate residuals for a solved Poisson regression

Usage

## S3 method for class 'logitmodel'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  link = "odds",
  pearson = FALSE,
  deviance = FALSE,
  ...
)
## S3 method for class 'logitmodel'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  link = "odds",
  pearson = FALSE,
  deviance = FALSE,
  ...
)

Arguments

x

result model for a regression, class logitmodel

df

a data.table containing the columns of interest

control

list of parameters controlling the convergence, see the control options vignette for details

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

link

pearson

boolean to calculate pearson residuals

deviance

boolean to calculate deviance residuals

...

can include the named entries for the assign_control list parameter

Value

returns matrices of residuals

Calculates the Residuals for a completed logistic model

Description

Residual.logitres uses user provided data, person-year/event columns, vectors specifying the model, and options to calculate residuals for a solved Poisson regression

Usage

## S3 method for class 'logitres'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  pearson = FALSE,
  deviance = FALSE,
  ...
)
## S3 method for class 'logitres'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  pearson = FALSE,
  deviance = FALSE,
  ...
)

Arguments

x

result object from a regression, class logitres

df

a data.table containing the columns of interest

control

list of parameters controlling the convergence, see the control options vignette for details

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

pearson

boolean to calculate pearson residuals

deviance

boolean to calculate deviance residuals

...

can include the named entries for the assign_control list parameter

Value

returns matrices of residuals

Calculates the Residuals for a poisson formula

Description

Residual.poismodel uses user provided data, person-year/event columns, vectors specifying the model, and options to calculate residuals for a solved Poisson regression

Usage

## S3 method for class 'poismodel'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  pearson = FALSE,
  deviance = FALSE,
  ...
)
## S3 method for class 'poismodel'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  pearson = FALSE,
  deviance = FALSE,
  ...
)

Arguments

x

result object from a poisson model, class poismodel

df

a data.table containing the columns of interest

control

list of parameters controlling the convergence, see the control options vignette for details

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

pearson

boolean to calculate pearson residuals

deviance

boolean to calculate deviance residuals

...

can include the named entries for the assign_control list parameter

Value

returns matrices of residuals

Calculates the Residuals for a completed poisson model

Description

Residual.poisres uses user provided data, person-year/event columns, vectors specifying the model, and options to calculate residuals for a solved Poisson regression

Usage

## S3 method for class 'poisres'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  pearson = FALSE,
  deviance = FALSE,
  ...
)
## S3 method for class 'poisres'
Residual(
  x,
  df,
  control = list(),
  a_n = NULL,
  pearson = FALSE,
  deviance = FALSE,
  ...
)

Arguments

x

result object from a regression, class poisres

df

a data.table containing the columns of interest

control

list of parameters controlling the convergence, see the control options vignette for details

a_n

list of initial parameter values, used to determine the number of parameters. May be either a list of vectors or a single vector.

pearson

boolean to calculate pearson residuals

deviance

boolean to calculate deviance residuals

...

can include the named entries for the assign_control list parameter

Value

returns matrices of residuals

Checks OS, compilers, and OMP

Description

System_Version checks OS, default R c++ compiler, and if OMP is enabled

Usage

System_Version()
System_Version()

Value

returns a list of results

Examples

System_Version()
System_Version()

Automates creating a date since a reference column

Description

Time_Since generates a new dataframe with a column containing time since a reference in a given unit

Usage

Time_Since(df, dcol0, tref, col_name, units = "days")
Time_Since(df, dcol0, tref, col_name, units = "days")

Arguments

df

a data.table containing the columns of interest

dcol0

list of ending month, day, and year

tref

reference time in date format

col_name

vector of new column names

units

time unit to use

Value

returns the updated dataframe

Examples

library(data.table)
m0 <- c(1, 1, 2, 2)
m1 <- c(2, 2, 3, 3)
d0 <- c(1, 2, 3, 4)
d1 <- c(6, 7, 8, 9)
y0 <- c(1990, 1991, 1997, 1998)
y1 <- c(2001, 2003, 2005, 2006)
df <- data.table(
  m0 = m0, m1 = m1,
  d0 = d0, d1 = d1,
  y0 = y0, y1 = y1
)
tref <- strptime("3-22-1997", format = "%m-%d-%Y", tz = "UTC")
df <- Time_Since(df, c("m1", "d1", "y1"), tref, "date_since")

library(data.table)
m0 <- c(1, 1, 2, 2)
m1 <- c(2, 2, 3, 3)
d0 <- c(1, 2, 3, 4)
d1 <- c(6, 7, 8, 9)
y0 <- c(1990, 1991, 1997, 1998)
y1 <- c(2001, 2003, 2005, 2006)
df <- data.table(
  m0 = m0, m1 = m1,
  d0 = d0, d1 = d1,
  y0 = y0, y1 = y1
)
tref <- strptime("3-22-1997", format = "%m-%d-%Y", tz = "UTC")
df <- Time_Since(df, c("m1", "d1", "y1"), tref, "date_since")

Package 'Colossus'

Help Index

Fully runs a case-control regression model, returning the model and results

Description

Usage

Arguments

Value

Examples

Fully runs a cox or fine-gray regression model, returning the model and results

Description

Usage

Arguments

Value

See Also

Examples

Fully runs a cox or fine-gray regression model with multiple column realizations, returning the model and results

Description

Usage

Arguments

Value

See Also

Examples

Automates creating a date difference column

Description

Usage

Arguments

Value

See Also

Examples

uses a table, list of categories, and list of event summaries to generate person-count tables

Description

Usage

Arguments

Value

See Also

Examples

uses a table, list of categories, list of summaries, list of events, and person-year information to generate person-time tables

Description

Usage

Arguments

Value

See Also

Examples

Generic background/excess event calculation function

Description

Usage

Arguments

Value

See Also

Predicts how many events are due to baseline vs excess for a completed poisson model

Description

Usage

Arguments

Value

See Also

Predicts how many events are due to baseline vs excess for a completed poisson likelihood boundary regression

Description

Usage

Arguments

Value

See Also

Splits a parameter into factors

Description

Usage

Arguments

Value

See Also

Examples

Applies time dependence to parameters

Description

Usage

Arguments

Value

See Also

Examples

Interprets a Colossus formula and makes necessary changes to data

Description

Usage

Arguments

Value