Optimized recommendation engine

A recommendation system, comprising:

• a database;
• and
• a recommendation engine configured to:
• perform offline training of a first model at regular intervals based on data in the database;
• based on a request from a user:
• determine results for the request using the offline-trained first model,
• perform online training of a second model and determine results for the request using the online-trained second model, and
• identify common result items that separately occur in both the results determined using the offline-trained first model and the results determined using the online-trained second model;
• provide recommendations based on the common result items;
• and
• measure the accuracy of the recommendations.

A computer-program product comprising computer-executable instructions embodied in one or more non-transitory computer-readable media that, when executed, causes a computing system to perform a method for providing recommendations, the method comprising:

• providing a plurality of data variables representing characteristics of customers;
• analyzing the data variables by performing a correlation analysis;
• providing a data model of the analyzed data variables;
• based at least in part on the data model, performing offline training of a first model at regular intervals using an artificial neural network (ANN);
• responsive to a request from a user:
• determining results for the request using the offline-trained first model,
• performing online training of a second model using a support vector machine (SVM) and determining results for the request using the online-trained second model, and
• identifying common result items that occur in both the results determined using the offline-trained first model and the results determined using the online-trained second model;
• and
• based on the common result items, generating a potential customer list;
• and
• providing a recommendation based on the potential customer list.