change model registry version number

[salelkafrawy]

• Updated the model version number
• Fixed the best XGBoost model we saved in Q5 (xgboost_lasso instead of xgboost_SHAP)

We need to do:

• make sure the right columns are passed to the updated xgboost_lass (the columns can be found in the blog post)
• add xgboost_feats_non_corr as our best performer so far
• add xgboost_SMOTE and lr_SMOTE to the evaluation (just to see how the test set will perform)

[TimkLee]

Update: Problem encountered with xgboost_lass , used xgboost_SHAP instead. Replaced NN with xgboost_feats_non_corr .

SMOTE performance comparison: