Prediction of Rice Using Random Forest Algorithm with Geographical Information System (GIS)

Farmers’ decision-making processes could change if information technology is used to the agricultural industry. In order to anticipate future yield, the researchers in this study used a GIS-based model, automated learning by machine approaches, and extracting valuable insights from datasets. The training set and test data for yield status were built on information acquired from the provincial agriculture. A spatial map showed where the future yield would be at its highest and lowest levels. A prediction model is created in the first stage. In the second stage, a system that estimates future rice yields will be developed utilizing time series analysis and a GIS-based model for visualization. To effectively complete the data mining process, the proponents employed the cross-industry standard method for data mining. An attribute evaluator was used to process the data after it had been preprocessed using WEKA’s attribute selection filtering. The twelve abiotic agricultural parameters are climate, humidity level, time of year, elevation, water supply, growing date, amount of land, output per area, growing method, precipitation, sunlight exposure, and wind velocity. The five biotic agricultural variables are soil, saline, fertilizer, rice variety, and insect/pest diseases. The study offers a structure that can be utilized to improve the application of data mining in agriculture. A computed kappa value for the algorithm used to determine whether rice yield status as “high” or “low” was 0.7327, it is significant to recommend that the set of rules was trustworthy. The Provincial Agriculture Office may establish, incorporate, manage, oversee, and carry out all strategies, programs, initiatives, and actions related to the findings of this study.