Cascade Parallel Random Forest Algorithm for Predicting Rice Diseases in Big Data Analysis

Abstract

Experts in agriculture have conducted considerable work on rice plant protection. However, in-depth exploration of the plant disease problem has not been performed. In this paper, we find the trend of rice diseases by using the cascade parallel random forest (CPRF) algorithm on the basis of relevant data analysis in the recent 20 years. To confront the problems of high dimensions and imbalanced data distributions in agricultural data. The proposed method diminishes the dimensions and the negative effect of imbalanced data by cascading several random forests. For experimental evaluation, we utilize the Spark platform to analyze botanic data from several provinces of China in the past 20 years. Results for the CPRF model of plant diseases that affect rice yield, as well as results for samples by using random forest, CRF, and Spark-MLRF are presented, and the accuracy of CPRF is 96.253%, which is higher than that of the other algorithms. These results indicate that the CPRF and the utilization of big data analysis are beneficial in solving the problem of plant diseases.