Home-Work Detection

Home-work detection is an inference technique that identifies a mobile device's most probable home location and work location by analyzing the spatial and temporal regularity of its observed positions over weeks or months. It exploits the fact that most people spend predictable periods at home (nighttime and weekends) and at work (weekday business hours).

Algorithms cluster a device's stay points and weight them by time-of-day and day-of-week. The location where a device is most frequently observed during nighttime hours (e.g., 9 PM–7 AM) on weekdays and throughout weekends is inferred as the home location. The location most frequently visited during daytime working hours (e.g., 9 AM–5 PM) on weekdays, excluding the home cluster, is inferred as the workplace. Confidence thresholds ensure that only devices with sufficient observation density and temporal regularity receive labels. Common refinements exclude vacation periods, filter out commercial venues misidentified as residences, and handle multi-location workers.

Commute analysis relies on home-work detection to map origin-destination flows and measure average commute distances and times. Foot traffic analytics use inferred home locations to estimate the geographic origin of a store's visitors, answering the question: where do our customers live? Census validation and population estimation studies use aggregated home detection as a complement to traditional enumeration. Urban planners study jobs-housing balance and the geographic mismatch between where people live and work. Home-work detection transforms anonymous mobility traces into the anchor-point framework that enables commute studies, visitor origin analysis, and population-level spatial understanding.