Surrogate model for time-series model interpretation

A method comprising:

• storing a plurality of models that perform time-series machine learning based on a plurality of different algorithms, respectively;
• executing the plurality of models based on test data and determining accuracy values and interpretability toughness values for the plurality of models;
• selecting a most accurate model from among the plurality of models as a core model based on the accuracy values and selecting a most interpretable model from among other models remaining in the plurality of models as a surrogate model based on the interpretability toughness values;
• building a composite model comprising the core model, the surrogate model, and instructions for generating a debrief model for debriefing the core model by combining a trend component and a seasonal component from the surrogate model with a residual component from the core model to generate the debrief model;
• generating the debrief model, based on the instructions, by combining the trend component and the seasonal component from the surrogate model with the residual component from the core model;
• executing the generated debrief model to determine variable contributions of each of a plurality of variables of one or more of the plurality of models;
• initiating presentation of the determined variable contributions on a user interface;
• and
• storing the composite model within memory.

A non-transitory computer-readable medium comprising instructions which when executed by a processor cause a computer to perform a method comprising:

• storing a plurality of models that perform time-series machine learning based on a plurality of different algorithms, respectively;
• executing the plurality of models based on test data and determining accuracy values and interpretability toughness values for the plurality of models;
• selecting a most accurate model from among the plurality of models as a core model based on the accuracy values and selecting a most interpretable model as a surrogate model from among other models remaining in the plurality of models as a surrogate model based on the interpretability toughness values;
• building a composite model comprising the core model, the surrogate model, and instructions for generating a debrief model for debriefing the core model by combining a trend component and a seasonal component from the surrogate model with a residual component from the core model to generate the debrief model;
• generating the debrief model, based on the instructions, by combining the trend component and the seasonal component from the surrogate model with the residual component from the core model;
• executing the generated debrief model to determine variable contributions of each of a plurality of variables of one or more of the plurality of models;
• initiating presentation of the determined variable contributions on a user interface;
• and
• storing the composite model within memory.