facebook-archive
facebook-archive copied to clipboard
kaustubhhiware
Where have you been?
- What are the places frequently visited by you?
- How often do you visit the same place?
- how does it correlate with the location of you house/ college? Did you notice a pattern here?
Just to be clear, this issue is different from #9 since that issue focuses solely on a cluster map. Any analysis related to location should be discussed on this issue.
If this issue is unassigned, I would like to take it up. I think we can do this by plotting a scatter plot of Place visited vs distance from school/college, and keeping the size of the point proportional to the frequency of visiting the place
@purvaudai just a suggestion: once you start working on this, you might want to sort the top 10 places, with the most frequently visited on the left. this can give an added clear insight on how the distance of location maps to the frequency of visiting :)
I was wondering if I could use geopy library to get the location name from the coordinates given in the facebook json, because a plot with the coordinates instead of the places name, would not seem meaningful
@purvaudai the concern you've raised about is relevant to #9, not this issue. Please go through issue #9 and read this issue again. #9 deals with a cluster map and visualization, this issue only deals with analysis part.
If you're interested you can take up that issue after this.
So, are you are trying to say that I should let the x-axis of the plot be marked with coordinates, because the json file only has information in coordinates, which wont make much sense to the user.
Oh, I'm so sorry I misunderstood.
For the clubbing part, yes, feel free to use anything you like. It won't make sense to have coordinates in the x-axis, names would be preferable. You can club any two places in the vicinity of 10 km (or whatever you find appropriate) as one place.
So, even if you live in a large city like Delhi, the possible locations in Delhi you can have is restricted to 5 / 10 / depending on the data. The clubbing part (only street names and city) should be easily available in some pip modules, you might not even have to code it.
If anything I'm saying isn't clear, let me know.
I dont understand how will I show the correlation between home and other places. Please explain this point.
I worked up from scratch and got geopy running over the list of coordinates, so after iterating through the locations we can find the occurrences of the places in that list. What more do we have to consider?
@Animesh-Chouhan there's the issue of you using your phone in multiple locations within your home, but those being marked as different locations. The initial discussion was related to how we can resolve that.
However, if you were to say, use your phone in your own home and use your phone in you friends home that's atleast 100m away, these locations should be treated differently.
If you have worked upon something, and have some results to share, do post them here or personally via pinging me on Gmail or gitter. We can work out the details and how to proceed with this issue accordingly.
So what I did is to first get the address out of the coordinates and then look out for the locations which are invariably distinct and then get the respective coordinates and then proceed to plotting or further processing.
Marking as assigned. Currently being worked upon by @animesh-chouhan