
PREPARING THE DATA FOR ANALYSIS
VARIABLE SELECTION
We have selected only the statistically relevant variables for our model building purpose.
-
Variables which show non-significant results in t-test and chi-square tests are not considered for model building. Non significant p values reflects that these variables do not have any effect on attrition.
-
The variable EmployeeCount has only one value 1 and hence it is not a very useful variable four our study and hence we have dropped it from model building.
-
The variable Over18 has only one value "Y" signifying that all every employee in the organization is above 18 years old. However this variable is not vary useful since it takes on only one value.
-
The variable StandardHours takes on only one vaue 80, i.e. 80 hours, which is the standard number of working hours for every employee per week and hence is fixed. Again being fixed this variable is of no use to us.