Comparison of life patterns from mobile data in Bangladesh

Digital economy is growing worldwide with spreading internet and mobile devices. The possibilities are discussed to apply these accumulated data of mobile devices for monitoring purposes, because more detailed data could be collected on activities or status of individuals more efficiently and frequently. Existing studies show some specific features or behavior or mobility patterns can be detected with mobile phone data. However, classifications or clustering of people behavior patterns are not attempted, though such classification may be useful in understanding “human landscape” and help design better sampling for field surveys to collect more detailed data for in-depth analysis and to correct biases due to biased mobile phone holdings and usage. We analyzed human behavior in Bangladesh using “Call Detail Records” of mobile devices for classifying people life patters and understanding the patterns in relation with income index and proximity to cities etc. For this purpose, we calculated features on estimated dwelling areas and subscriber's trajectory. Through comparing the description of “life behavior” patterns from CDR to physical environment and official census, we found that the differences on the distribution of “life behavior” patterns on area and the efficiency varied from the duration of the data.