I am trying to improve classification of imbalanced dataset creditcard fraud using SMOTE imbalanced_learn. But, in this it generates the data to 50%, can we give a specific number for the data to be generated? I want to track the classifier performance with the increase in generated data. Any help will be highly appreciated.
$\begingroup$
$\endgroup$
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
SMOTE is an algorithm for generating as many minority samples as you like. Thus, you can generate as many samples as you want.
$\begingroup$
$\endgroup$
Some of the packages such the one in python (imbalanced-learn) allows you to set the balancing ratio (which is 1 in the case of 50% minority and 50% majorities).
If you are not using a package without such a ratio parameter, you may use the index of the newly generated samples (if unavailable compare the input to the output and use only the delta) to find the new samples and then randomly chose the number of new desired samples only out of them.
