Empirical validation of website quality using statistical and machine learning methods
Poonam Dhiman, Anjali Anjali · 2014
The analysis of quantitative measure of large set of websites plays a significant role in evaluating the quality of websites. The paper, computes 22 metrics using a tool developed in MATLAB. Website quality prediction is developed using statistical and some machine learning methods. The work has been validated using dataset collected from webby awards web site. The results are analyzed using Area Under the Curve (AUC) obtained from Receiver Operating Characteristics (ROC) analysis. The results show that the model predicted using the random forest and Bayes Net methods outperformed over all the other models. Hence, based on these results it is reasonable to claim that quality models have a significant relevance with design metrics and the machine learning methods have a comparable performance with statistical methods. Univariate analysis results provide an empirical view for website design guidance and suggest which metrics are more important for website development.