A Regression Model for Analysis of Bounce Rate using Web Analytics

Abstract

Bounce rate is an effective parameter to measure the quality of any website. Bounce rate refers to the percentage of visitors that leave a website (or “bounce” back to the search results or referring website) after viewing only one page a website. High bounce rate is bad as it depicts that the content on a site didn’t match what the visitor was looking for so he left without viewing another page. Since bounce rate equates to visitors taking absolutely no action on a website so this metric could be used as a measure of success .This paper analyses the bounce rate of a website based on web analytics data. In this paper, analysis of bounce rate will be based on performance of website. Data is collected using Google Analytics tool. After applying preprocessing techniques to data an eleven step regression model is built using the various attributes like Average Server Response Time, Average Server Connection Time, Average Redirection Time, Average Page Download Time, Average Domain Lookup time and Average Page Load Time. Mathematical equation is constructed on the basis of outcome of result so that bounce rate can be analyzed and predicted. Model is further refined after establishing the correlation between various attributes. Correlation is established to improve the accuracy in analysis and prediction of bounce rate. This regression model gives insight about the various parameters involved and their effect on bounce rate. Qunatile Quantile plot is constructed to see if plausible data is normally distributed. This complete experiment is done using R Studio