Optimizing machine learning model total payment volume predictions using model stacking

A method for improved machine learning modeling, the method comprising:

• obtaining, by a computer system, total payment volume data corresponding to a past period of time for an entity system configured to process electronic transactions;
• inputting, by the computer system, the total payment volume data into a forecasting model that is configured to predict first total payment volumes for future periods of time based at least in part on historical total payment volume data;
• outputting, by the computer system using the forecasting model, a first prediction of a total payment volume for a future period of time;
• acquiring, by the computer system, prediction enhancing data corresponding to the past period of time from one or more computer systems external to the entity system;
• inputting, by the computer system, the first prediction and the acquired prediction enhancing data into a machine learning model trained, using at least (a) the historical total payment volume data (b) historical predicted first total payment volumes outputted from the forecasting model and (c) historical prediction enhancing data, to predict second total payment volumes for the future periods of time, wherein the machine learning model is a different model from the forecasting model;
• outputting, by the computer system using the machine learning model, a second prediction of the total payment volume for the future period of time;
• comparing, by the computer system, the second prediction of the total payment volume to a real-time total payment volume for the entity system;
• and
• based on a detected difference between the real-time total payment volume and the second prediction of the total payment volume, causing, by the computer system, the entity system to adjust a server traffic capacity to accommodate a predicted peak payments per time interval for processing the electronic transactions.

A non-transitory machine-readable medium having stored thereon machine-readable instructions executable to cause a computer system to perform operations comprising:

• obtaining total payment volume data corresponding to a past period of time for an entity system configured to process electronic transactions;
• inputting the total payment volume data into a forecasting model that is configured to predict first total payment volumes for future periods of time based at least in part on historical total payment volume data;
• outputting, using the forecasting model, a first prediction of a total payment volume for a future period of time;
• fetching prediction enhancing data corresponding to the past period of time from one or more databases;
• inputting the first prediction and the acquired prediction enhancing data into a machine learning model distinct from the forecasting model and trained, using at least (a) the historical total payment volume data, (b) historical predicted first total payment volumes outputted from the forecasting model, and (c) historical prediction enhancing data, to predict second total payment volumes for the future periods of time;
• outputting, using the machine learning model, a second prediction of the total payment volume for the future period of time;
• comparing the second prediction of the total payment volume to a real-time total payment volume for the entity system;
• and
• based on a detected difference between the real-time total payment volume and the second prediction of the total payment volume, adjusting a server traffic capacity of the entity system to accommodate a predicted peak payments per time interval for processing the electronic transactions.