An Experimental Analysis of Crop Yield Prediction using Modified Deep Learning Strategy
P. Shyamala Bharathi, V. Amudha, G. Ramkumar, T. J. Nagalakshmi, N. Nalini, P. Jagadeesh
Abstract
Agriculture is the most significant and important backbone to the country's economy and as compare to other countries Indian civilization depends this agricultural field a lot. Different climate conditions such as rainfall, temperature, humidity levels, pesticide problems and so on need to be monitored continuously to maintain the agricultural field in good manner. Now-a-days, there are lots of Artificial Intelligence assisted technologies are available to predict the climate conditions and report it properly to the respective user. In this paper, a novel deep learning strategy is designed to support agricultural field to predict the crop yield level in fine manner, in which the proposed learning scheme is called as Modified Deep Learning Strategy (MDLS). This MDLS is derived from the conventional learning schemes called K-Nearest Neighbor and the Decision Tree Algorithms. The proposed approach consider the parameters such as rainfall ratio, pesticide usage and the weather conditions like temperature level as the prediction constraints to analyze the crop yield nature. The resulting section shows the proper efficiency ratio of all the mentioned algorithms in clear manner with graphical representations. A novel crop yield prediction dataset is considered to estimate the prediction level of crops, in which it is obtained from the open source database called Kaggle. The performance evaluation of the proposed approach is portrayed in the resulting section as well as it is cross-validated with the conventional learning schemes called k-Nearest Neighbor and the Decision Tree algorithms to prove the efficiency of the proposed approach called Modified Deep Learning Strategy.