Product Engineering Review: Financial Fraud Detection InDB

[ForbiddenDevil]

Reviewer 1 comments:

• Warnings should not be ignored.
• Use of matplotlib.pyplot instead of teradataml plot()
• Use of glm_prediction.result.to_pandas() should be avoided
• Use of sklearn.metrics mean_absolute_error, roc_auc_score, roc_curve teradataml ROC() returns AUC, FPR, TPR. These can be used to plot ROC curve using teradataml plot()

Reviewer 1 suggestions:

• Use teradataml plot
• use teradataml ROC(), TDGLMPredict()

Reviewer 2 comments:

• In section 5: Use of ScaleTransform
• In section7: Use of TDGLMPredict

Reviewer 2 suggestions:

• One can make use of sf_fit object to use ScaleFitTransform as sf_fit.transform.
• Instead of TDGLMPredict, glm_model.predict() can be used.

[ForbiddenDevil]

Incorporated suggested changes. PR: https://github.com/Teradata/jupyter-demos/pull/576

Reviewer 1 comments:

• Warnings should not be ignored. --- This is done to keep notebook clean. It is standard across all the notebooks.
• Use of matplotlib.pyplot instead of teradataml plot() --- Fixed in pass6
• Use of glm_prediction.result.to_pandas() should be avoided --- Fixed in pass6
• Use of sklearn.metrics mean_absolute_error, roc_auc_score, roc_curve teradataml ROC() returns AUC, FPR, TPR. These can be used to plot ROC curve using teradataml plot() --- Fixed in pass6

Reviewer 1 suggestions:

• Use teradataml plot --- Fixed in pass6
• use teradataml ROC(), TDGLMPredict() --- Fixed in pass6

Reviewer 2 comments:

• In section 5: Use of ScaleTransform --- Correction made
• In section7: Use of TDGLMPredict --- Correction made

Reviewer 2 suggestions:

• One can make use of sf_fit object to use ScaleFitTransform as sffit.transform. --- Thank you for the input. Changes made accordingly_
• Instead of TDGLMPredict, glm_model.predict() can be used. --- Thank you for the input. Changes made accordingly

[ForbiddenDevil]

Merged. Can be moved to "Done" phase.