ffctn hackons la-corruption

Post on 07-May-2015

670 Views

Category:

Documents

1 Downloads

Preview:

Click to see full reader

TRANSCRIPT

ffunctioninc.

© FFUNCTION INC, 2011

ETUDE DE CAS

VISUALISER DESDONNÉES OUVERTES

Sébastien Pierre, FFunction inc.@Hackons la Corruption., Novembre 2012

www.ffctn.com

ffunctioninc.

© FFUNCTION INC, 2011

INFOGRAPHIC : SE7EN SUMMITS

ffunctioninc.

© FFUNCTION INC, 2011

GOOGE DATAVIZ CHALLENGE 2010 (FINALIST)

ffunctioninc.

© FFUNCTION INC, 2011

NATIONAL GEOGRAPHIC SOCIETY'S PROJECTS

ffunctioninc.

© FFUNCTION INC, 2011

2008Canadian Federal

Travel & HospitalityExpenses

ffunctioninc.

© FFUNCTION INC, 2011

SOME THINGS HAVEN'T CHANGED SINCE 2008

➔ SCRAPING DATAin the absence of open-data, journalists will often be in the same context, having to spend time to collect, explore and assess the quality of the data.

ffunctioninc.

© FFUNCTION INC, 2011

SOME THINGS HAVEN'T CHANGED SINCE 2008

➔ FROM DATA TO STORYEach dataset is a discovery, getting a (compelling) story out of it is still a major challenge.

ffunctioninc.

© FFUNCTION INC, 2011

THE DATA

ffunctioninc.

© FFUNCTION INC, 2011

As the result of a federal government directive*, Travel and Hospitality Expenses have been published on the web in Canada since 2004

* Called “proactive disclosure”

ffunctioninc.

© FFUNCTION INC, 2011

http://www.tbs-sct.gc.ca/pd-dp/gr-rg/index-eng.asp

ffunctioninc.

© FFUNCTION INC, 2011

Data is (still) not directly accessible, and hosted on each specific ministry website, in a specific format.

ffunctioninc.

© FFUNCTION INC, 2011

ffunctioninc.

© FFUNCTION INC, 2011

NON-OPEN DATA

ACCURACY PROBLEMS

DATA MAY BE MISSING

DATA NOT UP TO DATE

ffunctioninc.

© FFUNCTION INC, 2011

22Mb SQL filescraped by citizens(available on Github)

ffunctioninc.

© FFUNCTION INC, 2011

A DATASET WHICH TURNS OUT TO BE A BIT OPAQUE...

ffunctioninc.

© FFUNCTION INC, 2011

BUILDING A TOOL TO EXPLORE THE DATA

ffunctioninc.

© FFUNCTION INC, 2011

Basic analysis of the data

ffunctioninc.

© FFUNCTION INC, 2011

Thinking about how to represent the data

ffunctioninc.

© FFUNCTION INC, 2011

Thinking about the flow of interaction

ffunctioninc.

© FFUNCTION INC, 2011

Importing and visualizing the data

ffunctioninc.

© FFUNCTION INC, 2011

Mapping out the different types of expenses (travel, hospitality & guidelines)

ffunctioninc.

© FFUNCTION INC, 2011

Simplifying the representation (expenses vs guidelines, over guidelines is in red)

ffunctioninc.

© FFUNCTION INC, 2011

Changing the focus (under/over guidelines instead of total spending)

ffunctioninc.

© FFUNCTION INC, 2011

Adding guides to improve reading the information

ffunctioninc.

© FFUNCTION INC, 2011

Adding filtering to narrow down to subsets of the data

ffunctioninc.

© FFUNCTION INC, 2011

Trying alternative representations on the data

ffunctioninc.

© FFUNCTION INC, 2011

Trying even more alternative representations on the data

ffunctioninc.

© FFUNCTION INC, 2011

THE RESULThttp://ffctn.com/a/expensevisualizer

ffunctioninc.

© FFUNCTION INC, 2011

ffunctioninc.

© FFUNCTION INC, 2011

ffunctioninc.

© FFUNCTION INC, 2011

I just found out the 5 top spendingFederal depts, check it out at http://ur1.ca/a3spt

ffunctioninc.

© FFUNCTION INC, 2011

ffunctioninc.

© FFUNCTION INC, 2011

ffunctioninc.

© FFUNCTION INC, 2011

1

FINDINGS

ffunctioninc.

© FFUNCTION INC, 2011

TRENDS ONLY BECOME APPARENTWITH THE PROPER MODE OF REPRESENTATION

Cumulative spending

Monthly spending

ffunctioninc.

© FFUNCTION INC, 2011

PROBLEMS IN THE DATA QUALITYBECOME VISIBLE

ffunctioninc.

© FFUNCTION INC, 2011

Spending of ministers for all departments

THINGS YOU WOULD EXPECTARE NOT NECESSARILY THERE

ffunctioninc.

© FFUNCTION INC, 2011

DATA TO STORY: CHALLENGES

➔ NON-OPEN DATA– Missing or incomplete data: is the problem in the

scraper or in the actual data?– At least you now have a tool to assess (and improve)

the data quality

ffunctioninc.

© FFUNCTION INC, 2011

DATA TO STORY: CHALLENGES

➔ NOT WHAT I THOUGHT– You might expect something about the data,

but the visualization might prove your wrong– You might have been looking for something specific

but you cannot see it in the visualization

See my “30 min of data visualization”workshop for more on this...

ffunctioninc.

© FFUNCTION INC, 2011

DATA TO STORY: CHALLENGES

➔ DID I TRY HARD ENOUGH?– There's no secret: you'll find something interesting if

you explore your data enough.– If everything fails, you can at least get fun facts or

controversial examples out of it.

ffunctioninc.

© FFUNCTION INC, 2011

HOSPITALITY EXPENSES SKYROCKET IN 2008 !!

ffunctioninc.

© FFUNCTION INC, 2011

2.5xAs much

INDUSTRY CANADA'S BIG SPENDER

MINISTER DIRECTOR

ffunctioninc.

© FFUNCTION INC, 2011

WAR IS COSTING CANADA AN ARM AND A LEG!

3 MILLIONS!(over a period of five years)

ffunctioninc.

© FFUNCTION INC, 2011

WE NEED OPEN DATA!

THIS IS NOT AN ACCEPTABLE PROACTIVE DISCLOSURE!

THE BEST STORY IS:

ffunctioninc.

© FFUNCTION INC, 2011

1

TOOLS

ffunctioninc.

© FFUNCTION INC, 2011

https://github.com/OpenRefineGOOGLE REFINE

ffunctioninc.

© FFUNCTION INC, 2011

http://vis.stanford.edu/wrangler/DATA WRANGLER

ffunctioninc.

© FFUNCTION INC, 2011

http://datawrapper.de/DATA WRAPPER

ffunctioninc.

© FFUNCTION INC, 2011

http://www-958.ibm.com/software/data/cognos/manyeyes/MANY EYES

ffunctioninc.

© FFUNCTION INC, 2011

R17 http://www.rseventeen.com/

ffunctioninc.

© FFUNCTION INC, 2011

THANKYOU!

sebastien@ffctn.com / @ffunction

WWW.FFCTN.COM

top related