Package 'quickReg'

A hypothetical dataset

Description

A hypothetical dataset extracted from package 'PredictABEL'

Usage

diabetes

Format

A data frame with 1000 rows and 14 variables:

• sex: 1=male, 2=female

• age: age of the participants(years)

• smoking: 0=non smoker, 1=smoker

• education: 0=without bachelor degree, 1=bachelor degree or above

• diabetes: diabetes mellitus, 0= health, 1= diabetes

• BMI: body mass index (kg/cm2)

• systolic: systolic blood pressure(mmHg)

• diastolic: diastolic blood pressure(mmHg)

• ... : other genetic information, see the ExampleData in PredictABEL-package .

See Also

ExampleData

---

Display a table used in paper

Description

Display count, frequency or mean, standard deviation and test of normality, etc.

Usage

display_table(data = NULL, variables = NULL, group = NULL,
  mean_or_median = "mean", addNA = TRUE, table_margin = 2,
  discrete_limit = 10, exclude_discrete = TRUE, save_to_file = NULL,
  normtest = NULL, fill_variable = FALSE)

display_table_group(data = NULL, variables = NULL, group = NULL,
  super_group = NULL, group_combine = FALSE, mean_or_median = "mean",
  addNA = TRUE, table_margin = 2, discrete_limit = 10,
  exclude_discrete = TRUE, normtest = NULL, fill_variable = FALSE)

Arguments

data	A data.frame
variables	Column indices or names of the variables in the dataset to display, the default columns are all the variables except group variable
group	Column indices or names of the first subgroup variables. Must provide.
mean_or_median	A character to specify mean or median to used for continuous variables, either "mean" or "median". The default is "mean"
addNA	Whether to include NA values in the table, see table for more details
table_margin	Index of generate margin for, see prop.table for more details
discrete_limit	Defining the minimal of unique value to display the variable as count and frequency, the default is 10
exclude_discrete	Logical, whether to exclude discrete variables with more unique values specified by discrete_limit
save_to_file	A character, containing file name or path
normtest	A character indicating test of normality, the default method is shapiro.test when sample size no more than 5000, otherwise lillie.test Kolmogorov-Smirnov is used, see package nortest for more methods.Use 'shapiro.test', 'lillie.test', 'ad.test', etc to specify methods.
fill_variable	A logical, whether to fill the variable column in result, the default is FALSE
super_group	Column indices or names of the further subgroup variables.
group_combine	A logical, subgroup analysis for combination of variables or for each variable. The default is FALSE (subgroup analysis for each variable)

Functions

• display_table_group: Allow more subgroup analysis, see the package vignette for more details

Note

The return table is a data.frame.

- P.value1 is ANOVA P value for continuous variables and chi-square test P value for discrete variables

- P.value2 is Kruskal-Wallis test P value for continuous variables and fisher test P value for discrete variables if expected counts less than 5

- normality is normality test P value for each group

Examples

## Not run: 
data(diabetes)
head(diabetes)
library(dplyr);library(rlang)
result_1<-diabetes %>%
 group_by(sex) %>%
 do(display_table(data=.,variables=c("age","smoking"),group="CFHrs2230199")) %>%
 ungroup()
result_2<-display_table_group(data=diabetes,variables=c("age","smoking"),
group="CFHrs2230199",super_group = "sex")
identical(result_1,result_2)
result_3<-display_table_group(data=diabetes,variables=c("age","education"),
group=c("smoking"),super_group = c("CFHrs2230199","sex"))
result_4<-display_table_group(data=diabetes,variables=c("age","education"),
group=c("smoking"),super_group = c("CFHrs2230199","sex"),group_combine=TRUE)

## End(Not run)

---

plot the result of regression result

Description

plot coefficients, OR or HR of regression models.

Usage

## S3 method for class 'reg'
plot(x, limits = c(NA, NA), sort = "order", title = NULL,
  remove = TRUE, ...)

plot_reg(x, limits = c(NA, NA), sort = "order", title = NULL,
  remove = TRUE, term = NULL, center = NULL, low = NULL, high = NULL,
  model = NULL, ...)

Arguments

x	A reg, reg_y or reg_x object without covariates information, 'cov_show=FALSE'
limits	A numeric vector of length two providing limits of the scale. Use NA to refer to the existing minimum or maximum value.
sort	A character determining the order of variables to plot, 'alphabetical' or 'order'. The later is the default to sort variables according to their effect size.
title	title of plot
remove	A logical, whether to remove infinite and NA value. The default is TRUE
...	additional arguments. When using your own regression results rather than from 'quickReg', please provide 'term','center','lower', 'high' and 'model' for plot.
term	A character of x axis variable in plot
center	A character of coefficient, OR or HR variable in plot
low	A character of lower confidence interval variable
high	A character of upper confidence interval variable
model	A character of model, "lm", "glm" or "coxph"

See Also

reg, reg_x, reg_y

Examples

reg_glm<-reg(data = diabetes, y = 5, factor = c(1, 3, 4), model = 'glm')
plot(reg_glm)
plot(reg_glm, limits = c(NA, 3))

---

Build regression models

Description

Build general linear model, logistic regression model, cox regression model with one or more dependent variables. Allow regression based on subgroup variables.

Usage

reg(data = NULL, x = NULL, y = NULL, group = NULL, cov = NULL,
  factors = NULL, model = NULL, time = NULL, cov_show = FALSE,
  confint_glm = "default", group_combine = FALSE)

Arguments

data	A data.frame to build the regression model.
x	Integer column indices or names of the variables to be included in univariate analysis. If NULL, the default columns are all the variables except 'y', 'group', 'time' and 'cov'.
y	Integer column indice or name of dependent variables, integer or character, allow more than one dependent variables
group	Integer column indice or name of subgroup variables.
cov	Integer column indices or name of covariate variables
factors	Integer column indices or names of variables to be treated as factor
model	regression model, see lm, glm, coxph for more details
time	Integer column indices or name of survival time, used in cox regression, see coxph for more details
cov_show	A logical, whether to create covariates result, default FALSE
confint_glm	A character, 'default' or 'profile'. The default method for 'glm' class to compute confidence intervals assumes asymptotic normality confint, you can also use profile likelihood method confint.glm, but it is pretty slow. In this case you could specify 'default' for speed.
group_combine	A logical, subgroup analysis for combination of group variables or each group variables. The default is FALSE (subgroup analysis for each group variable)

Value

The return result is a concentrated result in a data.frame.

---

Build regression models only one dependent variable

Description

Build general linear model, generalized linear model, cox regression model with only one dependent variables.

Usage

reg_x(data = NULL, x = NULL, y = NULL, cov = NULL, factors = NULL,
  model = NULL, time = NULL, cov_show = FALSE, detail_show = FALSE,
  confint_glm = "default", save_to_file = NULL)

Arguments

data	A data.frame to build the regression model.
x	Integer column indices or names of the variables to be included in univariate analysis. If NULL, the default columns are all the variables except 'y', 'time' and 'cov'.
y	Integer column indice or name of dependent variable, only one integer or character
cov	Integer column indices or name of covariate variables
factors	Integer column indices or names of variables to be treated as factor
model	regression model, see lm, glm, coxph for more details
time	Integer column indices or name of survival time, used in cox regression, see coxph for more details
cov_show	A logical, whether to create covariates result, default FALSE
detail_show	A logical, whether to create each regression result, default FALSE. If TRUE, with many regressions, the return result could be very large.
confint_glm	A character, 'default' or 'profile'. The default method for 'glm' class to compute confidence intervals assumes asymptotic normality confint, you can also use profile likelihood method confint.glm, but it is pretty slow. In this case you could specify 'default' for speed.
save_to_file	A character, containing file name or path

Value

If detail_show is TRUE, the return result is a list including two components, the first part is a detailed analysis result, the second part is a concentrated result in a data.frame. Otherwise, only return concentrated result in a data.frame.

Examples

reg_glm<-reg_x(data = diabetes, x = c(1:4, 6), y = 5, factors = c(1, 3, 4), model = 'glm')
##  other methods
fit<-reg_x(data = diabetes, x = c(1, 3:6), y = "age", factors = c(1, 3, 4), model = 'lm')
fit<-reg_x(data = diabetes, x = c( "sex","education","BMI"), y = "diabetes",
time ="age", factors = c("sex","smoking","education"), model = 'coxph')

---

Build regression models with more than one dependent variable

Description

Build general linear model, generalized linear model, cox regression model,etc.

Usage

reg_y(data = NULL, x = NULL, y = NULL, cov = NULL, factors = NULL,
  model = NULL, time = NULL, confint_glm = "default", cov_show = FALSE)

Arguments

data	A data.frame
x	Integer column indices or names of the variables to be included in univariate analysis, the default columns are all the variables except 'y' and 'time' and 'cov'.
y	Integer column indices or name of dependent variable
cov	Integer column indices or name of covariate variables
factors	Integer column indices or names of variables to be treated as factor
model	regression model, see lm, glm, coxph for more details
time	Integer column indices or name of survival time, used in cox regression, see coxph for more details
confint_glm	A character, 'default' or 'profile'. The default method for 'glm' class to compute confidence intervals assumes asymptotic normality confint, you can also use profile likelihood method confint.glm, but it is pretty slow. In this case you could specify 'default' for speed.
cov_show	A logical, whether to create covariates result, default FALSE

Value

The return result is a concentrated result in a data.frame.

---