Unravelling Crop Yield Secrets Through Identification of Significant Factors Using Machine Learning

The agriculture sector is currently dominated by machine learning technologies. Wheat in particular has traditionally been one of the key building blocks of the world food chain. Future food security will be dependent on increased output or production with higher yields. Hence a desideratum arises to incorporate approaches to identify factors affecting wheat yield. There has not been much effort done on the identification of factors in earlier studies. Therefore, to find the optimum factors, feature selection methods such as Chi-square, Correlation Heat map, Random Forest Classifier, Decision Tree Classifier, Extra Tree Classifier, XG boost, and Recursive Feature Elimination are applied in the proposed work. The dataset comprises the factors temperature, wind speed, precipitation, humidity, N, P, K, pressure, and yield. The component with the highest ranking relative to other dataset attributes is the best factor. Results showed that humidity is 37.95% important and temperature comes out to be 31.19% important. Both these factors become the major contributors to the accurate prediction of the yield of wheat crops. The proposed method used for appropriate feature identification has significant improvements and made it possible to use important characteristics to forecast crop yield, allowing the development of a better model to assist farmers in increasing their economic and social stability. Estimating the significance of the factors helps in the accurate prediction of wheat yield.