Continuous integration and automated testing of machine learning models

A system, comprising:

• a database comprising historical time series sales data, the historical time series sales data comprising aggregated sales data for one or more items sold in one or more stores over one or more past time periods, wherein the one or more stores maintain an inventory of one or more items at one or more stocking locations;
• and
• a server comprising a processor and memory and configured to:
• receive, from the database, the historical time series sales data comprising aggregated sales data for one or more items sold in one or more stores over one or more past time periods;
• train a first machine learning model to learn model parameters by identifying one or more causal factors that influence the sale of one or more items at one or more stores over one or more time periods, and to generate first predictions of the sales of one or more items in relation to one or more causal factors;
• train a second machine learning model, based on the first machine learning model, to generate second predictions of the sales of one or more items, wherein the second machine learning model is trained to test a new feature branch, wherein the new feature branch comprises one or more new causal factors, the one or more new causal factors each describing a strength of a factor variable contributing to the second predictions, and wherein the second machine learning model is trained using different chronological segments of the historical time series sales data;
• evaluate the predictions of the first and second machine learning models as compared to the historical time series sales data;
• and
• deploy the machine learning model that generated the predictions that are closer to the historical time series sales data to generate one or more subsequent predictions.

A non-transitory computer-readable storage medium embodied with software, the software when executed configured to:

• receive, with a server comprising a processor and memory, historical time series sales data comprising aggregated sales data for one or more items sold in one or more stores over one or more past time periods, wherein the one or more stores maintain an inventory of one or more items at one or more stocking locations;
• train a first machine learning model to learn model parameters by identifying one or more causal factors that influence the sale of one or more items at one or more stores over one or more time periods, and to generate first predictions of the sales of one or more items in relation to one or more causal factors;
• train a second machine learning model, based on the first machine learning model, to generate second predictions of the sales of one or more, wherein the second machine learning model is trained to test a new feature branch, wherein the new feature branch comprises one or more new causal factors, the one or more new causal factors each describing a strength of a factor variable contributing to the second predictions, and wherein the second machine learning model is trained using different chronological segments of the historical time series sales data;
• evaluate the predictions of the first and second machine learning models as compared to the historical time series sales data;
• and
• deploy the machine learning model that generated the predictions that are closer to the historical time series sales data to generate one or more subsequent predictions.