Monday, November 17, 2014

Homework for Session 5 - Batch 3 CBA

Hi all,

Please find here the individual homework for session 5.

Pls ensure you are able to replicate classwork examples with the R code sent before you try this one.

The idea is simple. I will require you to:

  • 1. Pull your facebook (FB) data. Your friends' list. Pls use the Rfacebook package and the instructions from the slides.
  • 2. Run the communities-detection algorithm on it.
  • 3. Paste a screenshot of the network with communities on a slide. Identify the top few clearly identified groups that you can see (like I'd shown for my FB pull in the class slides).
  • 4. Analyze the 5 largest communities you got in terms of (i) size, transitivity, density, centralities, and (ii) meaning (how does the community relate to the ego or focal person).

Submission format and deadline the same as in the past. Save your PPT as (your.full.name).pptx

Any queries etc, contact me.

Thanks.

Sudhir

12 comments:

  1. Hi Professor, I tried the Rfacebook package and followed instructions but was unable to get through. Found a link which says there is some problem with Rfacebook. please have a look.
    https://github.com/pablobarbera/Rfacebook/issues/5

    ReplyDelete
    Replies
    1. OK. Heard back from Aashish.

      You must, categorically, use only the 'unversioned' graph API. Mentioned clearly in the instructions in the PPT. If you use any versioned (2.1 or 2.2) API, it will show error.

      PS – Unversioned graph API will be available till December 2014 only.

      Sudhir

      Delete
  2. OK, I'll ask Aashish to take a look at it first. Meanwhile, pls use this link for now:

    http://www.wolframalpha.com/facebook/

    ReplyDelete
  3. Dear Prof,

    Greetings !!

    I was able to extract SNA from my facebook. Very informational and insightul. Thank you very much.

    Have few questions on centrality metrics
    1. There is a column that mention "cor". What does that mean
    2. wrt to eigen vector - how does it know which nodes are well connected people - did not feed any data on that aspect.

    Regards,
    Suryanarayana

    ReplyDelete
    Replies
    1. Hi SA,

      Pls use the help files for the associated functions. The arguments and the values for each func are typically well documented.

      I don't think we use 'cor' centrality anywhere, am not sure what it refers to.

      Sudhir

      Delete
  4. Hello Professor - I'm using gephi for facebook analysis as Rfacebook is not functional.
    http://www.datalicious.com/blog-post/facebook-data-visualisation

    Please confirm if this approach is fine. I'm able to analyse all the requirements of the assignment through gephi.

    Thanks,
    Hari

    ReplyDelete
    Replies
    1. Hi Hari Krishna,

      Yes, it is definitely OK with me. In fact, I think it merits wider dissemination in the class. Will ask Atreyee to email the class with this new tool. Thanks for sharing this discovery.

      Sudhir

      Delete
  5. Hello Sir,
    With reference to "4. Analyze the 5 largest communities you got in terms of (i) size, transitivity, density, centralities"
    what would define 'largest'? the top 5 largest communities in terms of size? In the anonymous data network-there are no clear clusters, so we can't make that out just by looking at the network graph, so we'll need to use a metric to decide which ones are larger. I am right now basing it on community.size.
    Thanks,
    Geetika.

    ReplyDelete
    Replies
    1. That's perfectly fine, Geetika.

      Sudhir

      Delete
    2. Sir how do I get communities data using
      http://www.wolframalpha.com/facebook/

      Regards
      Nitin

      Delete
  6. Hello Sir,

    I am working in IT industry for 3.5 years now and looking to move into analytics field.
    I am good at Excel and PL/SQL and Core Java programming.
    My Queries:
    1)Shall i enter analytics field
    2)Shall i opt for classes or self study is sufficient for R and SAS
    3)Roughly how much time do i spare for learning SAS/R both and individually
    4)What is the percentage of coding in Analytics field
    5)I heard self study is advisable but I have also heard getting good B-School name on your CV has edge over the former..
    shall i opt for some expensive certification from Premier B-Schools.

    Kindly Reply and Thank You in advance.

    -Kalpesh

    ReplyDelete
  7. Hello Alumni's, Prof,

    This is Rishu Gupta, CBA Batch 5, for past 15 days I was doing some RnD on this FB but till now no success :-(

    I read all your comments and tried the websites(especially, https://github.com/pablobarbera/Rfacebook/) but I guess FB is changing it's security settings at a fast pace.

    Please let me know if any one of you have tried connecting with FB in recent times, it will be great help.

    Also just for info, this is for my own RnD and CBA has not given any assignment on FB, so no rules/codes break. Just info :)

    Cheers,
    Rishu

    ReplyDelete