# How do I infer user activities from historical location data and other relevant data? [closed]

Suppose you have the historical location data of users, you also know the locations of shops on the street, can you infer what a user is doing (e.g. taking a tube, being on a business trip, traveling, staying at home, shopping)?

You can use any other obtainable data of the user, e.g. Twitter, Amazon, Expedia, Facebook...

We are doing local-based-service by profiling our users. It would be much appreciated if the admins can un-hold this question. We really need some informative inputs like the one provided by @Marcus D

Thanks so much!

• Is this homework? Here are some relevant papers: Location-Based Activity Recognition using Relational Markov Networks, The Hidden Permutation Model and Location-Based Activity Recognition, Inferring High-Level Behavior from Low-Level Sensors.
– Emre
Apr 15, 2016 at 15:48

It sounds a little like homework, your bio suggests otherwise, but geographic data is very interesting, so I thought I'd give it a bit of a comment!

If you have location tracking in Google turned on, which the data can be downloaded, you can determine many things.

You will need to enrich your data with extra GPS data sets of locations of schools, shops, particular buildings, places of worship,

Some for examples of what can be determined ...

• During a 'work day' you can determine a 'normal place of work'.
• In the 'evening' (for most people) the gps locator will become static, this is a good indication of 'home'
• You can probably infer if a person has children if they go to a primary or secondary school at 9am and 3pm. You can guess at their ages depending on the type of school.
• By following the route between 'work' and 'home' you can determine, by the congruence of roads or rail networks, and the locations of bus stops / train stations, the mode of transport.
• If the data you have enriched is detailed enough, you can determine an individual's choice in lunch options.
• If there are periods of increased velocity, depending on the time of day, you can determine that they perform exercise, either by running or by pushbike. Further to that you might be able to determine the consistency of their fitness resolve by the frequency and duration of their exercise.
• You can start building a more detailed profile of the person, by looking at how often they eat out, if they attend church, if they go to sporting matches, what team they support, where they get their car serviced. All these things can give insight into an individual with regard to their lifestyle, habits, preferences, affiliations.
• This is so informative. I need answers like this one! Apr 17, 2016 at 0:27