System and method for predicting food product sales utilizing a plurality of artificial-intelligence sub-models

A system for using a plurality of artificial intelligence (AI) sub-models to increase efficiency and explainability of an AI model trained based on segmented data for consumable products by bifurcating data dependencies, the system comprising:

• a server comprising a processor and a non-transitory computer-readable medium containing instructions that when executed by the processor causes the processor to perform operations comprising:
• receiving a request to generate a projected attribute indicating a market performance metric of a product within a future defined time period of the product using the plurality of AI sub-models;
• receiving one or more attributes of the product;
• executing the plurality of AI sub-models, comprising a first AI sub-model and a second AI sub-model to generate the projected attribute, the plurality of AI sub-models being trained by a process comprising:
• retrieving training data comprising historical products, training product attributes of the historical products, and corresponding market performance metrics;
• executing a data preparation protocol to generate a training dataset, wherein executing the data preparation protocol comprises:
• segmenting the training data according to categories of the historical products and corresponding departments;
• and
• generating, for each segment of the training data, a set of high impact training attributes representing ingredients that have a high impact on the corresponding market performance metrics based on determining a correlation between the training product attributes and the corresponding market performance metrics;
• and
• providing a filtered portion of the training dataset corresponding to the set of high impact training attributes to the plurality of AI sub-models to initiate training of the first AI sub-model and the second AI sub-model based on the filtered portion of the training dataset;
• and
• wherein executing the plurality of AI sub-models comprises bifurcating a generation of the projected attribute into two steps:
• a first step comprising executing, by the processor, the first AI sub-model to predict a distribution value based on at least some of the set of high impact training attributes, wherein the distribution value predicts how many retailers will carry the product based on the one or more attributes of the product;
• and
• a second step comprising executing, by the processor, the second AI sub-model to ingest the distribution value predicted by the first AI sub-model and predict the projected attribute based on the distribution value determined by the first AI sub-model and at least some of the set of high impact training attributes;
• and
• presenting the projected attribute.