Effects of using multi-category web pages on rank estimation of Google search engine results page.

This article focuses on improving search engine results page (SERP) rank estimation algorithms by employing machine learning models trained on datasets derived from multi-category web pages, rather than single-category ones. The study involved scraping Google SERPs using diverse English keywords across multiple categories, extracting on-page search engine optimization (SEO) factors with tools like ScreamingFrog, and preparing datasets for classification. Results demonstrated that classifiers trained on multi-category datasets achieved over 25% higher accuracy in predicting web page rankings compared to those trained on single-category datasets, with Gradient Boosted Trees identified as the most effective algorithm. Correlation analysis highlighted internal linking as the most influential on-page SEO factor affecting Google SERP rankings. The findings suggest that incorporating diverse web page categories in training data enhances the robustness and accuracy of SERP rank estimation models.