Forecasting the Yield of Thai Hom Mali 105 rice in Thung Kula Ronghai using Principal Component Panel Regression

The purpose of this research was to construct Principal Component Panel Regression to forecast and study
the factors of weather conditions that impact on agricultural production for yield of Dok Mali 105 rice variety in the Thung Kula Ronghai. The data used combines cross-sectional data and time series data, collecting from five provinces in the Thung Kula Ronghai area: Roi Et, Surin, Yasothon, Mahasarakham, and Si Sa Ket. Yearly rice production data from the Office of Agricultural Economics and monthly weather data in cultivating season (NASA/POWER) were gathered in each province from 2012 to 2021. The data was divided into two sets: a training set for model development and a test set for evaluating the model forecasting performance, using the Root Mean Square Error (RMSE).

It has been found that there was multicollinearity in weather variables therefore, the first 3 principal
components with 85.83% explain variance were constricted for panel regression. With principal panel regression models, the Time Fixed Effect Principal Panel Regression Model is the best model (RMSE= 18.33). In addition, the weather factor significantly effecting the yield of Dok Mali 105 rice variety is low wind speed in positive manner impact on the yield of Dok Mali 105 rice variety. The weather conditions affect rice yields differently each year.