I'm trying to produce week over week foot traffic comparison. Shall I use different core places crosswalk file for all time period?for each time period, or shall I just use the latest core places file for all time period

Questions & Help
,
1

Hi - I’m trying to produce week over week foot traffic comparison. Shall I use different core places crosswalk file for each time period, or shall I just use the latest core places file for all time period? Right now I have my code as: (1) read in foot traffic data for 2020 week 1 - latest week, (2) merge to Jan 2021 core places file, keep only matched. But I’m seeing the foot traffic in DC has dropped in the week of Jan.6, which doesn’t make sense…

2

You want to use the latest core places file; the core places file is improved and expanded over time, rather than churned, so the most recent one is always what you want.

As for the drop on the week of Jan 6, I’m not sure I entirely follow your method but it sounds like you might be comparing Jan 6 to the full year, which you wouldn’t want. Instead try constructing a time series of visits over time to DC POIs/aggregated DC, perhaps using visits_by_day.

3

Thanks Nick. As a follow-up question: I am interested in how many more visits DC has this month compared to last month. To get the comparable trend, is merging the patterns data with the latest core places file enough, or do I still need to normalize the visits?

4

You’ll still want to normalize the visits. I’d recommend using the total_devices_seen variable for DC (or nationally) in the normalization_data file to adjust each day’s visits before aggregating