User performance assessment with data mining approach: a case study

This paper presents how data mining techniques can be applied for evaluating and comparing performance of users of TESMA organizational performance management software with a reference user that is modeled automatically. Evaluation factors considered based on general user evaluation test results like “Number of characters typed”, “Download the attachments or not”, “time spent evaluating an indicator” and etc. Since logged user data is mixed with missing data and outliers, data cleaning plays a key role in statistical analysis and data modeling. We do a comparison on six different outlier detection methods in terms of accuracy. After preprocessing of data, using two different data mining modeling techniques user reference patterns are investigated and empirically assessed in terms of accuracy.

کلیدواژه ها:

Data Mining ، User Performance ، Data Preprocessing ، Reference Model.