Package moonBook

In moonBook: Functions and Datasets for the Book by Keon-Woong Moon

Function "mytable"

Function "mytable"" produce table for descriptive analysis easily. It is most useful to make table to describe baseline characteristics common in medical research papers.

Basic Usage

require(moonBook)
data(acs)
mytable(Dx~.,data=acs)

The first argument of function mytable is an object of class formula. Left side of ~ must contain the name of one grouping variable or two grouping variables in an additive way(e.g. sex+group~), and the right side of ~ must have variables in an additive way. . is allowed on the right side of formula which means all variables in the data.frame specified by the 2nd argument data. The sample data 'acs' containing demographic data and laboratory data of 857 patients with acute coronary syndrome(ACS). For more information about the data acs, type ?acs in your R console.

str(acs)

Explore a data.frame

You can use mytable function to explore a data.frame.

mytable(acs)

You can use formula without grouping variable(s).

mytable(~age+sex,data=acs)

Use of labelled data

You can use labels of the data. For example, you can add column labels and value labels of mtcars data using set_label() and set_labels() function of the sjlabelled package.

if(!require(sjlabelled)) {
    install.packages("sjlabelled")
    require(sjlabelled)
}
df=mtcars
df$am<-set_label(df$am,label="Transmission")
df$am<-set_labels(df$am,labels=c("automatic","manual"))
df$vs<-set_label(df$vs,label="Engine")
df$vs<-set_labels(df$vs,labels=c("V-shaped","straight"))

You can see the labels of this data.

str(df)

You can use the column label and value labels with mytable() function. The default values of use.labels and use.column.labels arguments of mytable() function are TRUE. You can change this arguments if you want.

mytable(df)
mytable(df,use.labels=FALSE, use.column.label=FALSE)
mytable(am~.,data=df)
# mytable(vs+am~.,data=df)

Choosing grouping variable(s) and row-variable(s)

You can choose the grouping variable(s) and row-variable(s) with the formula.

mytable(sex~age+Dx,data=acs)

You can choose row-variable(s) with . and + and - and variable name in an additive way.

mytable(am~.-hp-disp-cyl-carb-gear,data=mtcars)

Method for continuous variables

By default continuous variables are analyzed as normal-distributed and are described with mean and standard deviation. To change default options, you can use the method argument. Possible values of method argument are:

• 1: forces analysis as normal-distributed, default value
• 2: forces analysis as continuous non-normal
• 3: performs a Shapiro-Wilks test to decide between normal or non-normal

When continuous variables are analyzed as non-normal, they are described with median and interquartile range.

mytable(sex~height+weight+BMI,data=acs,method=3)

Because the method argument is selected as 3, a Shapiro-Wilk test normality test is used to decide if the variable is normal or non-normal distributed. Note that height and BMI was described as mean $\pm$ sd, whereas the weight was described as median and interquartile range.

choice of variable : categorical or continuous variable - my way

In many cases, categorical variables are usually coded as numeric. For example, many people usually code 0 and 1 instead of "No" and "Yes". Similarly, factor variables with three or four levels are coded 0/1/2 or 0/1/2/3. In many cases, if we analyze these variables as continuous variables, we are not able to get the right result. In mytable, variables with less than five unique values are treated as a categorical variables.

mytable(am~.,data=mtcars)

In mtcars data, all variables are expressed as numeric. But as you can see, cyl, vs and gear is treated as categorical variables. The carb variables has six unique values and treated as continuous variables. If you wanted the carb variable to be treated as categorical variable, you can changed the max.ylev argument.

mytable(am~carb,data=mtcars,max.ylev=6)

Combining tables

If you wanted to make two separate tables and combine into one table, mytable is the function of choice. For example, if you wanted to build separate table for female and male patients stratified by presence or absence of DM and combine it,

mytable(sex+DM~.,data=acs)

For more beautiful output : myhtml

If you want more beautiful table in your R markdown file, you can use myhtml function.

out=mytable(Dx~.,data=acs)
myhtml(out)
out1=mytable(sex+DM~.,data=acs)
myhtml(out1)

For more beautiful output : mylatex

If you want more beautiful table, you can use mylatex function.

mylatex(mytable(sex+DM~age+Dx,data=acs))

You can adjust font size of latex table by using parameter size from 1 to 10.

out=mytable(sex~age+Dx,data=acs)
for(i in c(3,5)) 
    mylatex(out,size=i,caption=paste("Table ",i,". Fontsize=",i,sep=""))

Export to csv file : mycsv

If you want to export your table into csv file format, you can use mycsv function.

mycsv(out,file="test.csv")
mycsv(out1,fil="test1.csv")

Following figure is a screen-shot in which test.csv and test1.csv files are opened with Numbers.

Use of ztable

You can use ztable() function from the ztable package.

require(ztable)
require(magrittr)
mytable(sex+DM~.,data=acs) %>%
    ztable %>%
    addSigColor %>%
    print(type="html")

densityplot

library(moonBook)
densityplot(age~sex,data=acs)
densityplot(age~Dx,data=acs)

Plot for odds ratios of a glm object

require(survival)
data(colon)
out1=glm(status~sex+age+rx+obstruct+node4,data=colon)
out2=glm(status~rx+node4,data=colon)
ORplot(out1,type=2,show.CI=TRUE,xlab="This is xlab",main="Odds Ratio")
ORplot(out2,type=1)
ORplot(out1,type=1,show.CI=TRUE,col=c("blue","red"))
ORplot(out1,type=4,show.CI=TRUE,sig.level=0.05)
ORplot(out1,type=1,show.CI=TRUE,main="Odds Ratio",sig.level=0.05,
        pch=1,cex=2,lwd=4,col=c("red","blue"))

For automation of cox's proportional hazard model

attach(colon)
colon$TS=Surv(time,status==1)
out=mycph(TS~.,data=colon)
out
HRplot(out,type=2,show.CI=TRUE,cex=2,sig=0.05,
       main="Hazard ratios of all individual variables")

moonBook documentation built on Jan. 5, 2022, 9:06 a.m.