SYSTEMS AND METHODS FOR IMPROVING THE INTERPRETABILITY AND TRANSPARENCY OF MACHINE LEARNING MODELS

摘要

Embodiments herein provide for a machine learning algorithm that generates models that are more interpretable and transparent than existing machine learning approaches. These embodiments identify, at a record level, the effect of individual input variables on the machine learning model. To provide those improvements, a reason code generator assigns monotonic relationships to a series of input variables, which are then incorporated into the machine learning algorithm as metadata. In some embodiments, the reason code generator creates records based on the monotonic relationships, which are used by the machine learning algorithm to generate predicted values. The reason code generator compares an original predicted value from the machine learning model to the predicted values from the machine learning model.