A Study on Castor Oil seed crop in Tamil Nadu Using Machine Learning and Non-Linear Mathematical Models

Castor is a non-edible industrial oilseed crop. Castor seeds are used for domestic, medicinal, and industrial purposes. Castor oil is used in machinery and particularly high-speed engines and airplanes. The present study has been undertaken to identify the best Non-Linear model and yield prediction model using machine learning techniques for castor oil seed crops in Tamil Nadu. 30 years data was collected from Season and crop Report of Tamil Nadu for Castor oil seed crop from 1990-2020. The best-fitted model was chosen based on model selection criteria like the highest coefficient of determination (R²), and with the least MAPE, RMSE, and MAE values. The four nonlinear models are fitted for the Area & Production of Castor oil seed crops. The results indicate that the Sinusoidal model is found to be the best fit for Area and Production since it has a high R² value (0.91) and low RMSE value. According to the yield prediction model for castor oil seed crop, the Machines Learning models such as the Random forest model, Logistic Regression model, and Support Vector classifier models are considered. The study indicates that the Random Forest model is found to be the best-fitted model based on model performance metrics and the Actual vs predicted graph also clearly indicate the coincidence between the actual and predicted yield. Using these fitted models one could be able to study the trend for the Area and Production of oilseed crops in Tamil Nadu.