mechkar (version 1.14.6)

train_test: Generates a training and test dataset

Description

This function generates a training and test datasets by randomly assigning individuals to each dataset.

Usage

train_test(data=NULL,train_name=NULL,test_name=NULL,prop=NULL,seed=123,tableone = FALSE)

Arguments

data

original dataset

train_name

a string that defines the name to be assifned to the train dataset object

test_name

a string that defines the name to be assifned to the test dataset object

prop

the proportion of the training dataset. The value is a fractional number between 0 and 1. The value default value is set to 0.6, indicating that the training dataset will contain 60% of the cases and the test dataset will contain the 40% of the cases.

seed

the desired seed. Using a constant seed value allows to obtain the same individuals on each group when running many times (important feature needed for replicability)

tableone

a logical value indicating if the Table1 function has to be generated for comparing the train and test division. Default is FALSE

Value

This function creates new variables using the names entered for the train and test partitions. Additionally, it returns the a table (based on the Table1 function) comparing all the available variables by partition. This helps understanding if the partition is balanced.

Examples

# NOT RUN {
### the following example will generate a train dataset named "train" which
### includes 70% of the records, while generating a test
### dataset called "test" and that includes 30% of the the original dataset.
df <- Theoph
df$Subject <- NULL
train_test(data=df,train_name="train",test_name="test",prop=0.7,seed=2)
# }