arpitg91 / performancescorecard

Get performance metrics and graphs of a scorecard

PerformanceScoreCard

Python script to automate the performance of scorecard. The script creates the charts for the following charts and profiles:

• ROC, AUC, KS, Score@KS
• Cumulative Gains Chart
• Precision Recall Chart
• Deciles Profile
• Confusion Matrix
• Histogram of classes
• Quadratic weighted kappa
• Linear weighted kappa
• Kappa
• Mean quadratic weighted kappa
• Weighted mean quadratic weighted kappa

Functions

The fuctions can be called from another python script to get individual performances.

get_roc(df,score,target,title,plot=1)

get_cum_gains(df,score,target,title)

get_precision_recall(df,score,target,title)

get_deciles_analysis(df,score,target)

confusion_matrix(rater_a, rater_b, min_rating=None, max_rating=None)

histogram(ratings, min_rating=None, max_rating=None)

quadratic_weighted_kappa(rater_a, rater_b, min_rating=None, max_rating=None)

linear_weighted_kappa(rater_a, rater_b, min_rating=None, max_rating=None)

kappa(rater_a, rater_b, min_rating=None, max_rating=None)

mean_quadratic_weighted_kappa(kappas, weights=None)

weighted_mean_quadratic_weighted_kappa(solution, submission)