Application of data mining techniques to predict internet usage consumption for personal objectives in the work place

Internet usage by employees for personal of inappropriate purposes can directly impact the productivity and efficiency of the organization. This translates to lost time, opportunity and money. In this research we use a data mining technique to build an internet usage consumption model by applying two different methods to web server log data: 1) decision trees based upon a C4.5 algorithm and 2) Multilayer perceptrons. The overall results obtained indicate that multilayer perceptrons with the cross validation have higher performance in classifying and predicting employee web browsing habits than decision trees. This data mining technique can therefore be a good candidate for helping organizations make more effective evaluation of their human and computer resources

员工出于私人用途或不当用途使用互联网，会直接对组织的生产力与运营效率造成负面影响，进而引发时间、机遇与经济成本的损失。本研究采用数据挖掘（Data Mining）技术，针对Web服务器日志（Web Server Log）数据应用两种不同方法构建互联网使用消费模型：1）基于C4.5算法（C4.5 Algorithm）的决策树（Decision Trees）；2）多层感知机（Multilayer Perceptrons）。整体实验结果表明，结合交叉验证（Cross Validation）的多层感知机在分类与预测员工网页浏览习惯方面，性能优于决策树。因此，该数据挖掘技术可作为助力组织更高效评估其人力与计算机资源的优质方案。