Performance Analysis of Novel Linear Regression Algorithm with Improved Accuracy Compared over K-Nearest Neighbor in Predicting Wind Power Generation

Computer Science and Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences,Saveetha University

Abstract

This work proposes a novel Linear Regression algorithm compared its performance with the K-Nearest Neighbor (KNN) algorithm for improving the accuracy of wind power generation prediction.In this study, two groups were created for the purpose of comparing the effectiveness of the KNN model (group 1) and the Linear Regression model (group 2) in predicting wind energy output. Each group consisted of 10 samples, resulting in a total of 20 samples used for the analysis. The data in this study were collected from an actual wind turbine and include the following factors: wind speed, altitude, humidity, air density, wind direction, and output power. The information was gathered at 10-minute intervals over the course of a year. The dataset was preprocessed, and the mean value of the corresponding variable was used to impute the missing values. Seventy percent of the data was used for training and thirty percent for testing. The training set was used to train the models, whilst the testing set was used to assess the effectiveness of the models. Python’s scikit-learn module was made use for the development of the Linear Regression technique. Based on statistical power (G-power) = 0.8, α = 0.05, CI of 95% confidence interval was also calculated. The observations indicate that the Linear Regression algorithm is more accurate than the KNN technique. The linear regression model achieved an accuracy of 82.15%, whereas the KNN model had a lower accuracy of 79.55% for predicting wind energy output. Additionally, the statistically significance values of the research was determined to be at a p-value of 0.001 (p<0.05). The algorithm was implemented and evaluated using real-world wind power generation data, and the findings demonstrate that, in terms of accuracy, This Linear Regression algorithm surpasses the KNN approach.