Suppose I have a dataset which has the columns date, time, latitude, longitude, place name visited by a user. I am wondering whether there is a way to write any algorithm to predict the path, the user would take at a particular time.

For example, I have the last two months date, time, latitude and longitude, can we predict the path the user would take at a particular date next month. I am not sure whether this is possible. But if any one can give me a idea of algorithm or any packages in R or python, it would be helpful. I can research on it.

  • $\begingroup$ I am doing research on this field to predict the user movement. I have read the four papers noted above. Where could I download some dataset about user movement? $\endgroup$ Apr 13 '18 at 4:19
  • $\begingroup$ Search for "gps trace dataset" and find results like this. Please take such questions to opendata.SE in the future. $\endgroup$
    – Emre
    Apr 13 '18 at 4:38

I think we've covered this question before but I can't find my response... A lot of work has been done in this area. I don't know of any packages, but the basic idea is to discretize the search space. One way is to just to use the cellular grid, which makes sense if you have that information. Another is to first extract the "locations" from the location trace; i.e., places the users revisits and spends significant time in. Then you can apply well-known techniques such as HMMs to estimate the next discrete location from the recent past. Most of the papers I read on this subject are about a decade old, so they came out before the deep learning flurry. If they were revisiting the subject I'm sure I'd see papers using recurrent neural networks.

Here are some papers for inspiration:

  • Learning and Inferring Transportation Routines
  • A personal route prediction system based on trajectory data mining
  • A Comparison of First- and Second-Order HMMs in the Task of Predicting the Next Locations of Mobile Individuals
  • Using GPS to Learn Significant Locations and Predict Movement Across Multiple Users

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.