Temporal aggregation effects in calculating typical spatial characteristics of typical human activities using trajectory data

Trajectory data is widely used to calculate spatial characteristics of human activities and to mine patterns of human behavior. When the time sampling intervals of the original trajectory data vary, the temporal observation scale of the moving object's location changes, which can affect the calculated human activity characteristics and analysis results, a phenomenon known as the temporal aggregation effect. Taking mobile phone data as a representative example, trajectory data is not specifically collected for human dynamics research; the varying time sampling intervals of the data result in widespread temporal aggregation effects. However, there is currently a lack of analysis on the temporal aggregation effects and their underlying mechanisms for typical human activity characteristics. To address this, this study utilized 123 days of intensive sampling mobile phone data from volunteers and selected six typical human activity spatial characteristics from four dimensions for analysis. The results show that: 1) There exists a temporal aggregation effect in calculating typical human activity spatial characteristics based on trajectory data. As the time sampling interval increases, the values of the indicators are generally underestimated, although the extent of underestimation varies across different indicators; 2) Indicators dependent on short-duration activities (e.g., daily travel distance, daily travel frequency, and daily travel spatial structure) are significantly affected, while those focusing on long-duration activities (e.g., maximum daily activity range, number of daily activity anchor points) are less affected. Comprehensive indicators (e.g., activity location entropy) are moderately affected; 3) When the sampling interval exceeds 30 minutes, the variation of indicators, except for daily travel distance, shows good consistency among individuals. These findings enhance the understanding of human activity spatial characteristics based on trajectory data and help improve the scientific foundation of decision-support systems.