Method, apparatus, and computer program product for predicting consumer behavior

A computer-implemented method, comprising:

• determining, using a processor, using a first machine learning model and for each cohort of a plurality of cohorts, a set of attributes of a plurality of attributes identified as having highest measures of importance for predicting a given metric relative to other attributes of the plurality of attributes, wherein the first machine learning model comprises a random forest model comprising a plurality of decision trees each comprising unique attributes of the plurality of attributes relative to other decision trees of the random forest model;
• identifying, using the processor, one or more first attributes of a plurality of attributes based on attributes associated with a first cohort of a plurality of cohorts to which a first consumer belongs;
• selecting, using the processor, a second trained machine learning model based on the first cohort, the second machine learning model comprising a plurality of decision trees generated based on the one or more first attributes associated with the first cohort, wherein each decision tree of the plurality of decision trees comprises unique attributes of the one or more first attributes relative to other decision trees of the plurality of decision trees;
• determining, using the processor and based on applying the second trained machine learning model to values specific to the first consumer for the one or more first attributes, a first prediction value;
• and
• transmitting, using the processor and to a first consumer device associated with the first consumer, an electronic communication comprising one or more promotions selected based at least in part on the first prediction value.

A non-transitory computer readable storage medium storing instructions that, when executed by at least one processor of an apparatus, cause the apparatus to:

• determine, using a first machine learning model and for each cohort of a plurality of cohorts, a set of attributes of a plurality of attributes identified as having highest measures of importance for predicting a given metric relative to other attributes of the plurality of attributes, wherein the first machine learning model comprises a random forest model comprising a plurality of decision trees each comprising unique attributes of the plurality of attributes relative to other decision trees of the random forest model;
• identify one or more first attributes of a plurality of attributes based on attributes associated with a first cohort of a plurality of cohorts to which a first consumer belongs;
• select a second trained machine learning model based on the first cohort, the second machine learning model comprising a plurality of decision trees generated based on the one or more first attributes associated with the first cohort, wherein each decision tree of the plurality of decision trees comprises unique attributes of the one or more first attributes relative to other decision trees of the plurality of decision trees;
• determine, based on applying the second trained machine learning model to values specific to the first consumer for the one or more first attributes, a first prediction value;
• and
• transmit, to a first consumer device associated with the first consumer, an electronic communication comprising one or more promotions selected based at least in part on the first prediction value.