Discussions

When analyzing POI duration patterns across years in our processed data, we observe a striking difference between 2023 and all other years. In 2023, approximately 21% of POIs appear for only 1 month of the year, 4% appear for 6 months, 0.5% for 9 months, and 58% for the full 12 months, creating what appears to be a realistic distribution of business persistence patterns.

However, for years 2019-2022 and 2024, virtually all POIs (97-100%) appear for exactly 12 months with almost no variation in duration patterns.

We have verified that our processing methodology is consistent across all years and that we have complete monthly file coverage for each year. We also notice that our 2023 aggregated data files are significantly larger (approximately 9.7GB) compared to other years (5-7GB each), suggesting a fundamental difference in the underlying data composition or volume.

We are wondering whether there was a methodological change in 2023 regarding POI inclusion criteria in your data collection process. This could relate to how new POIs are incorporated, minimum activity thresholds, or other data quality filters. Understanding whether this pattern reflects a change in your methodology would help us interpret our analytical approach correctly for our comparative research.

We would be grateful for any guidance you can provide about potential methodological changes or insights into interpreting these duration patterns across different years in your datasets.