Factor analysis on call detail record

We analyze the call detail record (CDR) data generated by mobile communication users. According to the traffic distributions produced by base stations, we take use of factor analysis method to elaborate the spatiotemporal characteristics. In this paper, we select eight base stations for subsequence study. Firstly, Pearson correlation coefficient and KMO indicator are used to decide whether the data is suitable for factor analysis. Secondly, we use principal component analysis (PCA) method to extract factors and only maintain three factors according to the eigenvalue of correlation matrix. Finally, combining the rotated factor loadings and factor scores, the three factors that we extracted can well explain the characters of business quarter, residential area and information technology (IT) zone in deferent time periods.