safenet

• 

  before bias mitigation for gender

• 

  after bias mitigation for gender

• 

  after bias mitigation

• 

  before bias mitigation for race

Bias mitigation for any CSV file, universal using LLM, getting correct columns that may exhibit bias and then using SMOTE ADASYN and in-processing mitigation using fair learn to mitigate bias. Results of before and after can be seen, you can change the file name from adult to your choice. the last column should be the target column.

End-to-end machine learning solution for bias removal in data and return a mitigated data frame.

Note: Works for any file and does in-process mitigation only for binary classification.

run: python remove_bias.py

(works for adult.csv)

Built With

• python