0
$\begingroup$

So I have a data set like so:

Email                  Posted           Received                            Platform
ss104@gmail.com 3/20/2012 1:11  Tue, 20 Mar 2012 01:07:05 -0700 (PDT)   Wikipedia
ss104@gmail.com 3/20/2012 1:11  Tue, 20 Mar 2012 01:07:22 -0700 (PDT)   Mailing list
ss104@gmail.com 3/20/2012 1:11  Tue, 20 Mar 2012 09:08:27 -0700 (PDT)   Twitter
ss104@gmail.com 3/20/2012 1:11  Tue, 20 Mar 2012 12:22:46 -0700 (PDT)   Wikipedia
ss104@gmail.com 3/20/2012 1:11  Tue, 20 Mar 2012 17:45:51 -0700 (PDT)   Mailing list
...

What this means is that a certain email address was posted at a certain time on a certain platform. It received an email at the time under Received column.

What I want to find is which platform sends the most amount of emails per week. (I'll be using python and matplotlib for visualizing). CPD (cumulative probability distribution) for this data is a good start.

I'm not sure how to start with this. Do I just count the occurrence of each platform, and divide it by the summation of difference in times (Received - Posted in weeks) for each email of that platform?

And then I could also make histogram or box plot for the data, but I'm not what actually to compute.

Can anyone give me pointers?

$\endgroup$
  • $\begingroup$ You have to subtract Received - Posted. Then group it by Platform. Now you have a bunch of numbers per platform. You can print a histogram, boxplot or estimated cdf for these bunches. But I have no idea how to do this in Python. $\endgroup$ – ziggystar Jul 19 '13 at 7:53
  • $\begingroup$ Yes, that's exactly what I thought. But I don't know how to go ahead with the cdf. Not with python just in general anyways. $\endgroup$ – KGo Jul 19 '13 at 8:35
  • $\begingroup$ Also, I think I would need something to normalize the values of weeks I get. Because the number of emails posted in not constant for each platform. $\endgroup$ – KGo Jul 19 '13 at 8:42
  • $\begingroup$ Could you explain how estimating the CDF (cumulative distribution function) would address your stated problem of finding the platform sending the most emails per week? (And what variable would the CDF apply to? That isn't at all apparent.) Why wouldn't just counting the emails per platform and finding the largest solve this problem? $\endgroup$ – whuber Jul 19 '13 at 14:16

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.