AI and Data Science Master the art of data science. Join now
Hi, I have been trying to improve the accuracy of my recall in XGBoost classifier. Currently, I am taking 1 year data and predicting for next 2 months. These time frames are curated according to the business. While running the XGB classifier, the accuracy that I am getting is 78%, precision is 69%, recall is 45% and F1 score is 52%. I feel recall of atleast 60% seems a good enough acceptable number. I have tried numerous methods as of now >
1) Re-engineered the feature set.
2) Hyper Parameter tuning - Randomized Search CV
3) selection of features to go into the model based on feature importance plot and shap value.
Nothing seems to work here. Can anyone please suggest any other idea to try out? Thank You.