Evaluation of influencing factors on tea production based on random forest regression and mean impact value

Abstract

Overproduction of tea in the major producing countries is an important factor which restricts the development of tea. Therefore, the factors from the economic, social and environmental system affecting tea production have become the focus of both academia and practice. Random forest regression (RFR) and mean impact value (MIV) were applied to evaluate the weights of variables. Firstly, RFR was preliminarily used to build a well-trained model, and then the weights of variables combining with MIV were calculated. Then, a well-trained model was constructed after variable selection to evaluate the importance of tea production from 2007 to 2016. The results revealed that the economic system and the social system are the main factors that affect tea production. The net production value and total population have little negative effects on tea production, while the area harvested has a little positive effect. Based on the research findings, governments and enterprises should develop and upgrade tea production technology, promote the exchange and cooperation in the international tea trade, then ultimately achieve sustainable development of the tea industry.