Toward Crops Prediction in Indonesia

Agriculture’s role and contributions in the modern era of globalization are crucially important. In order to meet the ever-increasing demands of the world’s population, agriculture has faced a variety of obstacles over the years. On the other hand, climate change will impact crop growth, yields, and agricultural production. Recently, numerous researchers have introduced numerous machine learning models to address problems in diverse fields, including agriculture. Advancements in machine learning and crop simulation modeling have created new opportunities to enhance agricultural prediction. These technologies have each provided distinctive capabilities and substantial improvements in prediction performance; however, they have been primarily evaluated separately, and there may be advantages to integrating them to further improve prediction accuracy. The purpose of this study is to forecast future crops in Indonesia based on temperature, precipitation, soil condition, and humidity dataset. This study utilizes Indian dataset, as Indonesian dataset is not yet available. The dataset consists of 2200 records from 22 plant species. We compare four machine learning algorithms which are decision tree, support vector machine, random forest, and K-nearest neighbor with accuracy as evaluation metric. The result shows that random forest model can predict the suitable crop to be planted on specific soil and weather conditions with the accuracy of 0.989697. The next step is to implement the model with Indonesian dataset. A plan optimization to choose the best set of plant species to produce or obtain will be our next challenge.