Deep Learning Based Bagged CNN with Whale Optimization Algorithm to Forecast the Productivity of Rice Crop using Soil Nutrients

Soil nutrients plays a major role in the growth of crop and yield of crop. Traditional methods in agriculture failed to produce high yield due to global warming and chemical fertilizers. Thus, efficient predictive model from machine learning can be used to produce better yield. There are two subdivisions namely Ensemble Learning (EL) and Deep Learning (DL) that performs better than the machine learning shallow methods. DL deals with models which are very complex and data that are non-linearly related whereas, EL methods can combine multi- base models to built efficient predictive models. DL with EL methods can combine the advantages of both methods and produces end model that shows great accuracy in performance. Also, ensemble methods, namely boosting and stacking helps to reduce the bias and Bagging called Bootstrap aggregating reduces the variance while using the different subsets of the same dataset. In this research, Bagged Convolutional Neural Networks(CNN) based prediction system has developed with WOA to forecast the productivity of Rice crop with the help of soil nutrients such as micro and macro nutrients along with pH value. This Bagged DLArchitecture reduces the overfitting of the data and produces an accuracy of around 85.8% with a error rate of about 14.2%.