چكيده به لاتين
Population growth in metropolises causes people presence, crowd congestion, and high density of urban trips in metropolitan zones. Economic consequences of this congestion include high fuel consumption, environmental pollution, and also citizen's time wasted in traffic. Therefore, hotspot detection and complimenting demand management methods would be highly effective in the negative consequences of congestion. Due to the importance of space and time among urban trips characteristics, spatiotemporal analysis is so useful in detecting trip behavior and urban activities pattern. Database as the most crucial part of this analysis, not only should provide a continuous spatiotemporal dataset, but also should represent the whole characteristics of the society.
Applying ITS-based datasets like mobile phone data would provide a low-cost dataset with a high penetration rate for spatiotemporal analysis. In the present study, we used big mobile data for Shiraz metropolitan. The preprocessing task on the dataset consists of data cleaning, data filtering (detecting outliers), and also stay point detection. Considering spatial unit preprocessing, we assigned the data from the base station point layer to the traffic analysis zones (TAZ). To do so, we calculated the area each TAZ shared the Voronoi cells and assigned the same share of detected activity in Voronoi to the target TAZ. Finally, data aggregation was done on the TAZs. Using area and population Per TAZ, we normalized the analysis variables. Moran’s I global and local analysis, exploratory time-series analysis, Mann-Kendall trend detection method, and also standard normal heterogeneity test have been used in the study. Considering two crucial dimensions of urban trips, we implemented Moran’s spatiotemporal pattern detection and spatial-temporal time series clustering using the Fourier function. City structure and hotspot detection were the additional implemented analysis. Results show a clustering structure in TAZs. Also increasing trend of activities, detected in time series analysis. The study also extracts the activity start time, morning peak, and evening peak of trips in the workday, semi-workday, and off-days. Spatiotemporal analysis results demonstrated the density changing pattern in 1-hour timeframes of all three days. As expected, hotspots were mainly detected in commercial, administrative, remedial, and higher-education land uses. On the other hand, zones with residential, recreational, and green spaces land uses were detected as hotspots on off days.
