Accuracy Improvement of Online Shopping Recommendation System with Natural Language Processing Using Novel XgBoost Classification Compared with Logistic Regression Method Based on Time

Online reviews are very important since they improve customer communication on a worldwide scale and impact consumer purchasing habits. Objectives apply Novel XGBoost Classification and Logistic Regression for the evaluation. The two algorithms that are used are Novel XGBoost Classification and Logistic Regression. In reality, this experiment only made use of 300 of the samples. Considerations include a pre-test G-power of 80%, a confidence interval of 95%, a threshold, and a sample size of 0.05%. By this work it is observed that the Novel XGBoost Classification update achieved 98.67% accuracy, whereas Logistic Regression just 50.58%. The data has major statistical analysis consequences. Note that the 2 tailed p-value is 0.000, below 0.05 which is statistically significant. From this work it is observed that the novel XGBoost Classification approach is superior than the Logistic Regression method in terms of accuracy.