Random Forest Time Series Forecasting

To predict the change in U.S. organic certification over time, the monthly certification change in certification was calculated as Certified - (Surrendered, Suspended, and Revoked) from the beginning of 2019, when this data appears to have been collected. A random forest model was created using Scikit Learn's RandomForestRegressor. In validation and testing, it was found that using two shifted monthly counts gave the best results. As part of the update processing for this application, the model is updated as well.

Prediction Outcomes



U.S. Monthly Certification Trends - Six Month Forecast

Below, the net change in certification is shown as Certified - (Surrendered, Suspended, and Revoked)


U.S. Certification Over Time - Six Month Forecast

Below, the estimated total count of certified operations in the U.S. over time is shown.