Title: Predicting web page performance level based on web page characteristics

Authors: Junzan Zhou; Yun Zhang; Bo Zhou; Shanping Li

Addresses: College of Computer Science and Technology, Zhejiang University, Hangzhou, China ' College of Computer Science and Technology, Zhejiang University, Hangzhou, China ' College of Computer Science and Technology, Zhejiang University, Hangzhou, China ' College of Computer Science and Technology, Zhejiang University, Hangzhou, China

Abstract: Large scale web applications based on complex architecture, spanning multiple data centres and content distribution networks, may face performance issue of high page load time. Recent surveys suggest two thirds of users encounter slow websites every week and that 49% of users will abandon a site or switch to a competitor. Web service providers can improve the performance at maintenance phase, but have few tools to predict the performance at early phases. In this paper, we present an approach to predict web page performance based on classification methods. Classification models are trained using known labels and it can predict user experience for a new page. We applied various classification techniques as predictors. We evaluate our solution on over 2,200 websites' landing pages. Experiments show that our framework can provide helpful prediction. Among the investigated classification algorithms, random forest achieves the best performance in terms of accuracy.

Keywords: user experience; statistical modelling; web page performance; performance prediction; web pages; page load time; logistic regression; websites; framework; classification; web page characteristics; web applications; website speed; landing pages; random forest.

DOI: 10.1504/IJWET.2015.072338

International Journal of Web Engineering and Technology, 2015 Vol.10 No.2, pp.152 - 169

Available online: 10 Jul 2015 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article