ltorgo
SMOTE creating floats from integers
Thank you for the time and effort you put into DWwR because it is incredibly useful. SMOTE has been especially useful.
Question Do you have a suggestion for how to stop SMOTE from creating floats from columns where all distinct values are all integers? i.e. input integers into SMOTE and SMOTE returns int's and floats.
Problem
For example, running the following creates floats from columns containing only integers:
dataset.bal <- SMOTE(target ~ ., dataset, perc.over=675, perc.under=100);
Input
#distinct values in column 124 dataset are all integers
levels(as.factor(dataset[,124]));
[1] "0" "4" "8" "12" "16" "20" "24" "28" "32" "36" "40" "44" "48" "52" "56" "60" "64" "68" "72" "76" "80" "84"
Output
#floats values have been added in column 124 in dataset.bal
levels(as.factor(dataset.bal[,124]));
[1] "0" "4" "8" "12""16" "17.4798878207803" [7] "17.8020473793149" "20""20.5099524259567" "20.6726439939812" "20.7490051034838" "21.7515262812376" [13] "23.4109582398087" "23.5898735374212" "24""24.0396314188838" "24.0953625235707" "24.148680685088" [19] "24.2740816418082" "24.3701336197555" "24.5072071170434" "24.5087074693292" "24.5124940760434" "24.7718890225515" etc..
By limiting (clearly telling the data type) the column type in the dataframe, SMOTE generates new values in integers.
It works for me: df['Column_name'] = df['Column_name'].astype(int)
