Optimizing Apple cultivation in West Bengal using Machine Learning

Agriculture in India, particularly in states like West Bengal, faces challenges such as declining productivity, limited adoption of modern techniques, and heavy reliance on imported crops. One significant opportunity lies in identifying optimal regions for cultivating crops like apples, which are currently imported in large quantities. Traditional methods of crop suitability assessment are often subjective and lack precision, resulting in suboptimal land use and increased dependency on imports. This paper addresses this issue by leveraging machine learning, particularly Random Forest classifiers, to analyze environmental factors like soil composition, temperature, pH, and rainfall, to predict the suitability of regions in West Bengal for apple cultivation. The methodology incorporates a balanced dataset through oversampling and under-sampling techniques, followed by feature selection to enhance model accuracy. The model was trained, evaluated, and optimized using hyperparameter tuning, achieving a strong performance with an accuracy, precision, recall, and F1-score of 0.75. The results demonstrate the model’s effectiveness in identifying hotspots for apple cultivation, offering actionable insights for improving local agricultural productivity. Recommendations for future work, including the integration of advanced data sources and optimization techniques to further enhance the model's accuracy and applicability have also been made in this paper.