Is there an annual version of the Patterns data?

Hi everyone. Is there an annual version of the Patterns data? We’re running into several cases for home_cbgs when the bottom truncation at <2 devices and bottom censoring at >2 but <4 devices distorts the analysis. An annual Patterns that aggregates monthly movement would reduce the number of these cases.

Hi @Alex_Zentefis_Yale_University, unfortunately, there is no pre aggregated annual patterns data - to my knowledge. But that could be something released in the future.

Until then, perhaps some normalization methods might help. What does your methodology look like thus far?

Okay, thanks @Jack_Lindsay_Kraken1. We’re thinking of applying some econometric techniques to handle the censoring and truncation. Do you know of other ways that researchers have dealt with the issue?

I will look into it and also defer this this @Nick_H-K_Seattle_University, he might be able to help more

@Alex_Zentefis_Yale_University, in July we will be adding home by census tract to pick up the long tail. is that helpful for your use case?

Tobit (which I imagine is what you’re referring to) is one way to go. Alternate options include replacing all the 4s with some lower number to reduce their pull, dropping them entirely, or aggregating to a higher, say county, level, but while reducing their pull in some way

Thank you @Nick_H-K_Seattle_University. That’s helpful!

@Lauren_Spiegel_SafeGraph can you elaborate? A variable of interest for us is home_cbgs for a POI. The long tail observations we’re missing because of the truncation at 2 devices and censoring at 4 devices. Will the home census tract variable added in July be at the POI level? And will it be backfilled in Patterns?

For each POI that has Patterns, just as now there is a visitor_home_cbgs column there will be a visitor_home_aggregation column which will be a dictionary like visitor_home_cbgs but keys will be census tract. We will be backfilling this.

Yes, I love to hear that

Great! Thank you, @Lauren_Spiegel_SafeGraph. We’re looking forward to the July release!