Code for Imbalance Learning With Imblearn and Smote Variants Libraries in Python Tutorial
View on Github
imb_learning.py
import numpy as np import pandas as pd from imblearn.under_sampling import RandomUnderSampler,EditedNearestNeighbours,InstanceHardnessThreshold,TomekLinks from imblearn.over_sampling import RandomOverSampler import smote_variants as sv df=pd.read_csv("creditcard.csv") y=df["Class"] X=df.drop(["Time","Class"],axis=1) print(y.value_counts()) under=RandomUnderSampler() X_und,y_und=under.fit_resample(X,y) print(len(X_und[X_und==1])==len(X_und[X_und==0])) over=RandomOverSampler() X_und,y_und=over.fit_resample(X,y) print(len(X_und[X_und==1])==len(X_und[X_und==0])) under_samp_models=[EditedNearestNeighbours(),InstanceHardnessThreshold(),TomekLinks()] for under_samp_model in under_samp_models: X_und,y_und=under_samp_model.fit_resample(X,y) svs=[sv.kmeans_SMOTE(),sv.Safe_Level_SMOTE(),sv.SMOTE_Cosine()] for over_sampler in svs: X_over_samp, y_over_samp= over_sampler.sample(X, y)
