A Durian Yield Prediction Method Based on an Improved Multiple Regression Model
Ruipeng Tang, Narendra Kumar Aridas, Mohamad Sofian Abu Talip · 2023
As a tropical fruit, many consumers like to eat durian, especially in Southeast Asia. However, durian production is affected by many factors, such as climate change, soil conditions, and irrigation volume. In order to help durian farmers to predict durian yield, this study proposes a durian yield prediction method based on an improved multiple regression model. This method introduces the residual principal component method to improve the traditional multiple regression algorithm and calculates the residual after a single regression. It also extract principal components from the original correlation factors and use the residual as a principal component for regression modeling. The error rates of the training and test set of the improved multiple regression model are reduced by 9.662% and 7.134% compared with before unimproved model, which show that the improved multiple regression model has greatly improved the fitting degree and prediction accuracy of yield compared with the traditional model, which is beneficial to durian growers in formulating effective planting plans based on yield fluctuations in different years. It can achieve better price competitiveness in the market.