Ant lion optimization in deep neural network for forecasting the rice crop yield based on soil nutrients

Agriculture is the backbone of the Indian economy; more over half of the world’s people depend on rice production. Environmental conditions, quality of soil, landscapes, and infestations of insects, water quality and accessibility, genotype as well as planning of the harvesting process are the primary predictors for the yield of crops. The growth of plants is significantly influenced by the soil’s quality. Essential nutrient deficiency or unavailability of soil nutrients might reduce the amount and quality of food produced. Decision Tree, Support Vector Machine, K-Nearest Neighbor as well as Naive Bayes are a few of the machine learning approaches created for monitoring the features of the soil and predicting the production for the rice crop. These techniques have impacts like low accuracy, high FPR, high computational time and cost. In order to forecast rice crop yield prediction effectively, Deep Neural Network (DNN) classifier by means of Ant Lion Optimization (ALO) is proposed and its results are analyzed and compared with the results of existing approaches. The two primary steps in the proposed method are data pre-processing as well as classification. There are a few missing values and outliers within the original dataset. Large Negative values are filled in the place of missing values replacement and min–max normalization is used to scale the dataset. The nutrient content of the soil is categorized by a DNN classifier utilizing ALO optimization, and rice crop yield predictions are made for a specific soil. In 100 epochs, the proposed method provides an improved accuracy of 86% having an error rate of roughly 14%. As a result, accurate rice yield output is projected that help farmers to get better rice crop yield in the future.