A Hybrid Model of Ensemble Method and Data Envelopment Analysis (DEA) for E-government Websites Efficiency Evaluation

Abstract

This paper examines the status of e-government portals in Jordan considering their performance measures. It involves data collection, data preprocessing, and Data Envelopment Analysis (DEA) of these websites. Some steps are added like data reduction through features selection and clustering. In this paper, we propose a new model for collecting data, analyzing data, classifying data, and evaluating the results. We have applied one type of DEA models which is BCC VRS model. Moreover, we implemented ensemble stacking algorithm to improve the accuracy of classifiers. The population of this research is 111 Jordanian e-government websites. According to DEA analysis, the number of inefficient websites is 30. The proposed model could be like a recommender system for web development community to consider the improvement of some factors of e-government websites. The results of ensemble method improved the results of these classifiers.