Anomaly in home_panel_summary

I’ll check for a concurrent increase in visits, but the dramatic jump in devices/visits still prevents us from publishing our research. Since we’re doing a year-to-year comparison, the jump in devices has a notice-able effect on trends we’re estimating. We’re specifically looking at how many students attended in-person instruction during the 2020-2021 school year.

Did Austin forward you the ZIP file I sent him? It contains the file showing the issue we’re experiencing, and the necessary notebooks, environment and files to reproduce the issue

Hey Jeff, thanks for the quick response. You’re right about the indexing approach, and we’re looking to use devices in the panel as a weight for visits. We tried normalizing the device panels, but we’re worried that the large increase in devices still inflate the average or rolling average.

Let me know when you take a look, and thanks again!

Thanks for taking a look! Typing up my reply right now