Application of Data Sampling for Crop Yield Prediction using Stochastic Gradient Descent Neural Networks

Crop yield estimation is a very important problem faced by our farmers. Various techniques have been used traditionally to determine the yield. Modern day algorithms, techniques like machine learning and neural networks have also been used previously to solve this problem. Historic data based on different parameters has been used to train the models and predict yield. Previous papers have considered the data as a whole to estimate results. In this paper we try to sample the data over different time intervals and granularise our parameters so that accurate results can be obtained. We use artificial neural networks over our sampled data to estimate the yield. Our results show a great improvement in the accuracy of prediction.