I have a data set that contains entities and lat/lon values for each one, potentially multiple lat/lon values.

I believe that some of these entities may actually be the same person, although I can't tell this from the data. Is there a method that can be used to determine the similarity between multiple entities based on the various lat/lon values each one has?

It would be easy if there was just one lat/lon value per entity, but I potentially have multiple. One approach would be to match entities that share all lat/lon values, but I still want to identify entities that partially match on lat/lon values (e.g. 2 out of 3 lat/lon values).

It seems like this might be a problem that has been solved before, so I figured I would ask. I'm honestly not even sure if I am thinking about this the right way.

  • $\begingroup$ What do you want to find out of this? Are you looking for entities who are in a similar distribution of locations, trying to match up entities who might have split into two different ones over time? Are there exact matches in the lat/lons (entities show up at a limited set of possible places) or are you just looking for ones that are nearby? $\endgroup$ – Danica Sep 26 '16 at 14:49
  • $\begingroup$ @Dougal I would like to find entities who are in a similar distribution over locations. The current assumption is that we could use some other information that we have to verify t hat the two entities are actually the same person, but with multiple records in our data set. $\endgroup$ – neelshiv Sep 26 '16 at 14:50
  • $\begingroup$ Okay, and are there a distinct number of possible locations or do you want to use a general distance between locations? $\endgroup$ – Danica Sep 26 '16 at 14:54
  • $\begingroup$ @Dougal We currently have raw lat/lon values. I think using distance makes sense, but we could also bucket lat/lon values into a grid and use the different portions of the grid as distinct locations. For what it's worth, users that visit multiple locations will probably be logging locations within the same city. I would like to identify people in the same city distinctly, so the grid would have to be very fine if I used that approach. $\endgroup$ – neelshiv Sep 26 '16 at 17:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.