Rice Blast (Magnaporthe oryzae) Occurrence Prediction and the Key Factor Sensitivity Analysis by Machine Learning

This study aimed to establish a machine learning (ML)-based rice blast predicting model decrease the appreciable losses based on short-term environment data. The average, highest and lowest air temperature, average relative humidity, soil temperature solar energy were selected for development. developed multilayer perceptron (MLP), support vector (SVM), Elman recurrent neural network (Elman RNN) probabilistic (PNN) evaluated by F-measures. Finally, sensitivity analysis (SA) was conducted factor importance assessment. result shows that PNN performed best with F-measure (β = 2) of 96.8%. SA in resulting main effect period is 10 days before happened. key factors found are minimum followed equaled maximum temperature. phase lag may cause lower dew point suitable pathogens growth. Through this study’s results, warnings can be issued advance, increasing response time farmers preparing related preventive measures, further reducing caused blast.