The 2nd Day of Analytics – Merge

Carrying on with our 12 days of Analytics highlighting our upcoming collaborative analysis features, I wanted to share with you what the Two Turtle Doves were really doing – Merge.

GeoCommons was founded on the principle of value in understanding the combination of disparate datasets for better intelligence. While the world is working on semantic technology of tomorrow, today we can use geography to combine data through common locations in the world. In particular, several of our customers initially reached out to us in search of a way to help them better identify their target markets based upon location and dynamic data they had collected from the area.

Last week we highlighted using GeoIQ to aggregate the dynamic social media mentions of brands through Twitter. We had several parallel searches specifically looking at mentions of Target and WalMart stores. What we wanted to see was the combined number of tweets and followers by State in order to see comparable social media market penetration.

In the recursive nature of “12 days of [Analytics]” I incorporated our new aggregation functionality to summarize the number of twitter messages and followers to US State. Doing this with both the Target and Walmart datasets, I then used the Merge capability to combine them. In this example, which you can see the map I’ve used our analytics to combine the two twitter streams together to get a comparative analysis and understand brand prevalence. This particular comparison only has limited insight, where it is clear that certain regions of the US are prevalent in their use of social media. In future posts we’ll look at the next steps in analytics to better understand the actual quantitative difference and trending.

Merge has a lot of interesting aspects and options. The most straight-forward is merely combining together two sets of data so that their combined locations are all in one dataset. This could be, for example, demographics data from multiple states to build a unified regional view, or the rivers, lakes, and ponds to build a common hydrology layer.

In combining two sets of data with similar geometries, the features are merged and this can just include all of the attributes from both – so in my case I have number of Target messages as well as number of Walmart messages. But Merge can also overwrite attributes. So for example I may have an old count of number of stores or users by state that when I merge in with another dataset may replace this number with the new data.

Through the flexibility of quickly combining data, users can easily see how multiple disparate datasets can be brought together – either through combining their information, or replacing old with new data.The same example could have combined social media data with store sales, crime data, or demographics.

You can download the resulting analysis from Black Friday Tweets From Mobile mentioning Target and Wal-Mart, by State – or see the map.