High Performance EDA and LDA Analysis: An Application for Wheat Yield Estimation

Abstract

A worldwide industry that provides food, business, and employment opportunities, agriculture is a key component of human life. Despite this, wheat is one of the most common armed crops and the production rate harms wheat yield every year. In this paper, a prediction method for wheat yield has been calculated with different environmental impact assessment parameters. Predictors of data are a predictive approach that helps to categorize the data based on the different grouping patterns. Exploratory data analysis (EDA) and Linear discriminant analysis (LDA) are very effective approaches for grouping the data. The main aim of this paper is to predict the wheat yield prediction through EDA, decision tree, random forest regressor, ensemble learning, and LDA to maximize accuracy. Different environmental impacts parameters such as average rainfall, average temperature, and pesticides have been used to predict the wheat yield. Also, ensemble learning has been used for the prediction and analysis of the model through the decision tree and random forest regressor. Moreover, the LDA has been used to classify the wheat yield dataset by applying a reduction approach of LDA. During wheat yield prediction, the decision tree achieves 0.025 losses in training time. Also, the performance of LDA and EDA has been calculated through squared error functions. During wheat yield prediction through EDA with environmental impact parameters, the Root means squared error (RMSE) is 18245.27 while the value of Mean absolute error (MAE) is 12334.75. Furthermore, the work of LDA has presented by supporting the data visualization through different graphs using pandas and Matplotlib library. This study provides the data reduction predictors approach to the wheat yield and explains the data-preprocessing technique used along with EDA and LDA for wheat yield prediction in different environmental impact parameters.