My question, is the relation between the POI and other visited brands directional? For example, 100 people visited Publix first and then 20% of them visited CVS that week.
Hi @debjanipriya - SafeGraph’s documentation goes into this a bit more: Weekly Patterns | SafeGraph Docs
These are the brands that the visitors to this POI also visit, on the same day that they visit the POI. The number mapped to each brand is an indicator of how highly correlated a POI is to a certain brand. The value is a simple percent of POI visitors that visited the other brand on the same day.
Same logic goes for same week brands.
It doesn’t sound like the order of the visit matters.
Hi, Thank you for your quick response. After your answer, I looked at my dataset and found some inconveniences. I am working on Florida’s weekly pattern data set and aggregating 44 weeks from 2020. If visiting order does not matter then the number of common visitors between any two brands should be the same. But I found different values. For example, in FL _Pattern_2020-01-06 main brand- Publix, Same week Brand-Walmart and visitor-3721 but when main brand-Walmart, same week brand-Publix then visitors was 6181. As I sum visitors from all the outlets so they should be equal. I found the same thing for other datasets as well. I appreciate your help.
hi @debjanipriya. Re-reading the SafeGraph docs, I think the important distinction is:
The value is a simple percent of POI visitors that visited the other brand on the same day.
It’s not a raw count, but a percent of visitors that went to other brands on the same day they visited one brand.
You could have the following scenario:
Brand A - Related Same Day Brand: 33 (Brand B), 33 (Brand C), 34 (Brand D)
Brand B - Related Same Day Brand: 50 (Brand A), 50 (Brand C)
In this example, A → B != B → A since the numbers above are percentages.
Hi , I am aware that values are in percentage. I converted them to the number of visitors. I attached sample data about what I did and the result. If there is no order common visitors should be equal when I aggregate them.Thank you
I think your data was just an example but the last line is incorrect. 30,000 x 3% = 900. Numbers still don’t add up those so it doesn’t answer the question.
I don’t have access to the data, but do the percentages of Related Brands always add up to 100%. The docs say they only show the first 20 brands so I don’t know if they calculate those 20 brands to add to 100%, or the 20 brands might not add up to 100%. The answer to this could cause a rounding discrepancy.