Systems and methods for providing machine learning model evaluation by using decomposition

A system comprising:

• memory storing textual information, a continuous model, a tree model, an ensemble model, and instructions;
• and
• at least one processor coupled to the memory and configured to execute the stored instructions to:
• generate a first decomposition for one or more continuous model outputs relative to a baseline population by using a gradient interface of the continuous model, a second decomposition for one or more tree model outputs relative to the baseline population by using tree structure information for the tree model, and a third decomposition for one or more ensemble model outputs relative to the baseline population by combining the first and second decompositions in accordance with ensemble information for the ensemble model, wherein the first and second decompositions are generated using one or more baseline input data sets representative of the baseline population;
• generate explanation information from the third decomposition using the textual information;
• and
• access from a modeling system:
• the gradient interface, the tree structure information, the ensemble information, input data sets of the baseline population, the continuous model outputs for each of the input data sets, the tree model outputs for each of the input data sets.

A non-transitory computer readable medium having stored thereon instructions comprising executable code that, when executed by one or more processors, causes the one or more processors to:

• generate a first decomposition for a continuous model output relative to a baseline population, a second decomposition for a tree model output relative to the baseline population by using tree structure information for a tree model, and a third decomposition for an ensemble model output relative to the baseline population by combining the first and second decompositions, wherein the first and second decompositions are generated using one or more baseline input data sets representative of the baseline population;
• select rejection explanation information for an observation input data set of the ensemble model based on the third decomposition, wherein the observation input data set being associated with a rejection;
• and
• output the rejection explanation information.