[ddj] Fwd: [Scoda-dev] Network analysis of landgrabbing data

Lucy Chambers lucy.chambers at okfn.org
Wed Dec 5 15:02:03 UTC 2012


Might anyone here be able to help with this? It's about Network
Analysis - which I know is used regularly by people tracking
companies, media neutrality etc.


---------- Forwarded message ----------
From: Tom Longley <tom at tacticaltech.org>
Date: Wed, Dec 5, 2012 at 6:48 AM
Subject: [Scoda-dev] Network analysis of landgrabbing data
To: scoda-dev at lists.okfn.org


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi there --

The plot so far: I've put together two gentle introductory courses as
part of Tactical Tech's work with the School of Data. The first, on
data cleaning; the second on descriptive analysis. Both use
spreadsheets. They're in the final run now. The data used in the
course is about large scale land acquisition in the developing world
by hedge funds for the biofuel industry
(http://datahub.io/dataset/grain-landgrab-data).

For the third course, I'm looking at at network analysis using a foss
tool called Gephi. With limited tinkering you make make neat things,
for example this showing the "land grabbing" transaction network

https://cloud.tacticaltech.org/public.php?service=files&file=%2Ftom%2Ffiles%2FSchool%20of%20Data%2Fexchanges.png

(the Gephi file for this is here:
https://cloud.tacticaltech.org/public.php?service=files&file=%2Ftom%2Ffiles%2FSchool%20of%20Data%2Fgrain_landgrab_20121205.gephi.zip)

*just remove the .zip extension - I added to this as our cloud
objected to .gephi as an extension*

Gephi certainly makes graphing more accessible, and I feel "graph
thinking" is such a useful basic skill in examining issues like this
one, and others involving companies and finance.

However, so far I have found the tool very challenging to use. This is
in part because of its byzantine UI, which can be overcome with a bit
of patience. However, the big issue is that its functions and the way
it behaves are unfathomable: it's hard to know what to expect if for
example you apply a Yifan-Hu Proportional algorithm  to a dataset. You
could apply it a 100 times and still not be any wiser about what it is
doing and why you'd choose it over a Force Atlas.

So I'm looking for some help: do you know anyone who can work with me
to be a tourist guide about network analysis, explaining the concepts
to me in plain language, and helping me de-mystify this for learners
at the School. I may be able to cover the cost of a day of someone's
time to help me out (confirmation pending on this).

Best regards,

Tom
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)

iQEcBAEBAgAGBQJQvzSLAAoJEMZAm28GQNw+kr0IALZiB4as8ZIUPA8n4WrsB+5I
xjWVpKDdk5+J6Gon82S0j2q/FCvnKS/kDchX7emDPBoAJxCCmozN+O1T+MASo15O
6FkcmRVy8jXhKUDXlP1+vO04JXK45Be57UgAh55mz0Td+XanZ/vkPPAtpPa7Il4m
zuC5/yKggwI8BCXA6PMzXst1LS3zeD8EAQeXTgvE6TQT1ker5Lsr0yAxgBWjkPoQ
c4cnc6HBE0+sj2NIyf3jbl4JRHXJB8ffMNJtuyNcqGgFRm1s12AhmpAeYKGr4jBh
APDZCVcSZiGbwxLSpwBbeXwDLGOiqRzrjN1GM/E13uBHnwz1rrR+I4V7MHtNdNI=
=IG8L
-----END PGP SIGNATURE-----

_______________________________________________
Scoda-dev mailing list
Scoda-dev at lists.okfn.org
http://lists.okfn.org/mailman/listinfo/scoda-dev
Unsubscribe: http://lists.okfn.org/mailman/options/scoda-dev


-- 
Lucy Chambers
Project Coordinator,
School of Data & OpenSpending
Open Knowledge Foundation
Skype: lucyfediachambers
Twitter: @lucyfedia




More information about the data-driven-journalism mailing list