Prediction of Grain Output in Anhui Province Based on Machine Learning

摘要

Food security has always been concerned by all countries. The grain production data of Anhui Province from 1990 to 2017 are selected, including: total grain production, sown area, etc. Low accuracy of food production forecasts has become a problem. We proposed a machine learning combination model GR-SVR based on Random Forest (RF), Gradient Boosting Decision Tree (GBDT) and Support Vector Regression Machine (SVR) to predict grain yield, using RF, GBDT, and SVR at the same time compare with GR-SVR. The results show that the GRSVR model has an accuracy rate higher than 96% in the prediction of grain output in Anhui Province, which is significantly better than GBDT, RF and SVR.