CROP YIELD PREDICTION SYSTEM USING K- NEAREST NEIGHBORS

Abstract

Agriculture is an integral part of economic development and food security in India, which has a majority of its population relying on agriculture for their livelihoods. But due to reasons such as climate change, global warming, unpredictable weather conditions, and other uncertainties in the environment, traditional methods of agriculture have been hampered to some extent resulting in low productivity and financial loss for farmers. In such a scenario, crop yield prediction becomes important for enhancing the performance of agricultural activities. The proposed thesis aims to develop a machine learning-based model for crop yield prediction through exploratory data analysis and K-nearest neighbors technique. An elaborate dataset covering various details such as the state-wise, district-wise, seasonal, and production details about crops grown in India has been explored to understand the important factors affecting the crop yield. The process of data preparation involves the use of a number of techniques for data cleaning, encoding, normalization, and feature analysis. An analysis was performed on several machine learning algorithms to test the efficacy of each algorithm for predicting crop yield. Among all the models examined, the KNN classifier outperformed the other algorithms due to its capability to accurately capture non-linear relationships in the data sets used in agriculture. The accuracy obtained from the generated model was 95.03%. This research shows that straightforward and efficient machine learning approaches can be successfully employed for crop yield prediction purposes. The designed approach may help farmers, agricultural professionals, and policymakers make better decisions regarding crop cultivation, resource allocation, and agricultural sustainability.