Prediction of Defective Products Using Logistic Regression Algorithm against Linear Regression Algorithm for Better Accuracy

The aim of the research investigation is being done to increase the precision and prediction of defective online products using management to find the defective online products using a machine, A novel logistic regression classifier is applied to the online products dataset that consists of 3197 records with 10 attributes and sample size=102. A framework for finding defective products in new defective online products logistics comparing novel logistic regression and linear regression algorithms has been developed. The accuracy and precision of the models were evaluated and recorded. The linear regression algorithm gives 76.0% accuracy in predicting the defective rate, whereas the novel logistic regression algorithm predicts the same with 84.2% accuracy. There exists a statistically different in a way that both (p=0.03; p0.05) Linear regression and logistic regression). The performance of novel logistic regression is significantly better and gives more appropriate results than the linear regression in finding defective rates for online products.