Utilizing Machine Learning for Forecasting Profitable Vegetable Crops and Land Use in Sri Lankan Agriculture

The agriculture sector faces numerous challenges, such as the selection of profitable vegetable crops, the reduction of surplus vegetable stocks, and the minimization of product losses in various vegetable varieties. This research study aims to address these challenges by employing machine learning to predict the most profitable vegetable crop and the corresponding land area required for cultivation. This anal-ysis considers multiple factors, including weather conditions, cultivation expenses, seasons, location, production data, and historical crop cultivation data. Multiple machine learning algorithms were employed to develop accurate crop prediction models namely, random forest RF, multinomial logistic regression MLR, and Long Short-Term Memory LSTM. The results indicate that multinomial logistic regression archives a higher accuracy of 0.84 for crop prediction and the random forest algorithm gives an accuracy of 0.83. The LSTM model demonstrated an R2score of 0.831. The model outputs the most profitable crop, which serves as input for the land extent prediction system. The application predicts the required land area for selected crops using RFR and LR. RFR gives a higher R2score of 0.956, indicating its superior fit to the data while the LR model achieved an R2score of 0.45. The root mean squared error of the RFR was 23.95 whereas the LR model had 89.29. The findings of this research study provide valuable information like crops and land extent to be cultivated in a particular area to increase productivity. The impact of this work extends to enhancing food security and reducing poverty in Sri Lanka by allowing farmers with the tools and knowledge to make correct decisions at the correct time.