I'm trying to build an agent-based model to study facebook usage. One important part is determining, at each step, whether a given user is online/active or not.

I am looking for a distribution of time spent online by percentage of users. I read in [this study][1] that this distribution is exponential, which is expected, but the study didn't provide any details of the function.

Just to be clear, I'm not looking for demographic breakdowns or anything. I'm looking for a function or dataset from which I can say "the top 5% of users spent 10 hours per day, the next 5% spent 6 hours", etc.

Not asking anyone to do my Googling for me, but I haven't found anything and I'd appreciate advise on where to look.

[1]:An Agent-Based Model of Message Propagation in the Facebook Electronic Social Network by Hamid Reza Nasrinpour, Marcia R. Friesen, and Robert D. McLeod, Member, IEEE. Published 2016, Accessed July 8 2017.

  • $\begingroup$ I am not sure what is your question and what are your expectations, what the answer should look like? I think that you will find function easy - sort data, calculate k-th percentile. For dataset, I do not think that it is on-topic here. $\endgroup$
    – Evil
    Commented Jul 7, 2017 at 21:52
  • $\begingroup$ Welcome to CS.SE! Does the paper mention the parameter of the exponential distribution, e.g., its mean? If not, have you tried contacting the authors of that paper, to ask them for the parameter of the exponential, and have you tried doing a literature search to look at other measurement papers on Facebook, to see if any of them provided any details of the parameters or distribution? $\endgroup$
    – D.W.
    Commented Jul 7, 2017 at 22:08
  • $\begingroup$ Where does it say that the time spent online is exponentially distributed? On a quick glance, it seems to assume the distribution is normal, not exponential (see Table I). And it's a simulation, so those seem to be assumptions rather than measurements. However it does cite other papers that make claims about the distribution. Have you read the prior papers? Seems like you could do more work here to read the paper carefully and summarize in the question what this paper says and what other papers it cites says. $\endgroup$
    – D.W.
    Commented Jul 7, 2017 at 22:14
  • $\begingroup$ Ok, @D.W. I have included a citation in the recommended format--sorry about that. The feedback I received indicated that people believed I may have misread and poorly summarized the cited paper, which called in to question the premise that the distribution was exponential, and that it may have been unclear what type of answer was being sought. I will take that advice and go back and reexamine the paper and other research to better specify my question. $\endgroup$ Commented Jul 8, 2017 at 5:59


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.