Systems and methods for alerting to model degradation based on distribution analysis

A detection modeling system comprising a processing device and a memory comprising instructions which are executed by the processing device for alerting to model degradation, the detection modeling system comprising a distribution analysis module configured to:

• select, by the processing device, model metrics for analysis, the model metrics being a measure of a parameter associated with a first analytical model;
• determine, by the processing device, normal distributions for model metric results for each of the selected model metrics, wherein each of the normal distributions comprises a plurality of values of model metrics, each of the plurality of values of model metrics from one of a plurality of analytical models, wherein the plurality of analytical models does not include the first analytical model;
• receive model metric values of the first analytical model for each of the selected model metrics;
• compare, by the processing device, the model metric values of the first analytical model to the normal distributions for model metric results for each of the received model metric values of the first analytical model;
• and
• alert, by the processing device, to model degradation of the first analytical model based on the comparison of the model metric values of the first analytical model to the normal distributions for model metric results.

A computer program product embodied in a computer readable storage medium comprising software that when executed by a processor performs a method comprising:

• selecting, by a processing device, model metrics for analysis, the model metrics being a measure of a parameter associated with a first analytical model;
• determining, by the processing device, normal distributions for model metric results for each of the selected model metrics, wherein each of the normal distributions comprises a plurality of values of model metrics, each of the plurality of values of model metrics from one of a plurality of analytical models, wherein the plurality of analytical models does not include the first analytical model;
• receiving model metric values of the first analytical model for each of the selected model metrics;
• comparing, by the processing device, the model metric values of the first analytical model to the normal distributions for model metric results for each of the received model metric values of the first analytical model;
• and
• alerting, by the processing device, to model degradation of the first analytical model based on the comparison of the model metric values of the first analytical model to the normal distributions for model metric results.