Make Your Dataset Representative: Fill Data Gaps with Active Measurements

Collecting real-world data in mobile communication has the drawback that these data are often not representing the real world. This happens due to passive collected data. The representativeness of a dataset can increase by adding active collected data. In this active data acquisition, we collect data with respect to specific characteristics like time of day or weekday, that are representing the integrity of the dataset. Thus, we explore a method to calculate our data gaps based on Active Learning and do active measurements to fill these gaps afterwards. Finally, we compare our newly collected data with our passively collected data to evaluate our method.