Indian Express covers Gramener at Open Data Unconference

Numbers with a view

How to give num­bers a pic­ture? A group of data col­lect­ors, data users and ana­lysts in Bangalore has an an­swer. Part of Open Data, they are at work “visu­al­ising” data on polit­ics ahead of the com­ing gen­er­al elec­tions.

Last month, at an Open Data “un­con­fer­ence”, a set of 32 slides ter­med ‘Visualising Politics’ was un­veiled and put on­line (http://www.slideshare.net/gramener/visualising-politics) as an ex­ample of how “see­ing” can make a dif­fer­ence in un­der­stand­ing cluttered tables that some­times only the trained eye can make sense of.

Sensing the in­terest in the sub­ject with cru­cial polls lined up, this group com­pris­ing pro­fes­sion­als from the private sec­tor as well as those as­so­ci­ated with the National Informatics Centre and Census be­lieves the data in the form they have presen­ted can help voters, and those look­ing for con­nec­tions between policies and how people vote, find some an­swers.

According to the group that first con­nec­ted on­line, this is of par­tic­u­lar im­port­ance in India given the sheer volume of data gen­er­ated in the coun­try, by the gov­ern­ment, the RTI ap­plic­a­tions and oth­er in­di­vidu­als.

The Bangalore ‘un­con­fer­ence’ was Open Data’s second an­nu­al meet­ing, and the key mover was Chief Data Scientist S Anand of Gramener. A Bangalore-based com­pany, Gramener has been grap­pling with how to make “big” data in a coun­try like India more ac­cess­ible and read­able, thereby al­low­ing it to be used more in­tel­li­gently and widely.

Says Anand: “We want to make slides that play with num­bers and tables avail­able any­way. But help­ing oth­ers visu­al­ise or see this as pic­tures makes it easy to as­sim­il­ate and use the data.”

Plotting data for largely the 2004 gen­er­al elec­tions, the group has ar­rived at some in­ter­est­ing con­clu­sions. For ex­ample, the per­cent­age of votes polled is con­sist­ently in­verse to the num­ber of con­test­ants in a race; and also, that the more densely pop­u­lated a con­stitu­ency, lower the vot­ing. Another in­ter­est­ing data shows that as the num­ber of con­test­ants in­creases, the per­cent­age mar­gin by which a win­ner wins in­creases.

If some of the group’s data shows dif­fer­ences in voter par­ti­cip­a­tion across re­gions, an­other brings out that Udaipur was the only con­stitu­ency in the coun­try where wo­men con­test­ants out­numbered men in the 2004 gen­er­al elec­tions; the men lost their de­pos­its.

The group is now work­ing on oth­er data, such as for­eign cur­rency re­ceived by NGOs, Karnataka Assembly res­ults from last time and Lok Sabha at­tend­ance, mapped party wise.

Analysts fa­mil­i­ar with gov­ern­ment data col­lec­tion have al­ways main­tained that India’s abil­ity to col­lect data for vari­ous things, across vast and in­ac­cess­ible re­gions and people, is com­mend­able. However, even in large ex­er­cises such as the Census, the data is col­lec­ted but without ne­ces­sar­ily seek­ing an­swers to any cor­rel­a­tions between vari­ables. So, sift­ing in the data, there is a tre­mend­ous scope to look for spe­cific an­swers to spe­cific ques­tions, which “visu­al­ising” helps make clear.

Several com­pan­ies and NGOs work­ing with Gramener use the data to plan the re­la­tion­ship between prob­lems and policies bet­ter.

Amrtha Kasturi Rangan, who works with Arghym, a found­a­tion work­ing on rur­al san­it­a­tion, says “visu­al­isa­tion” of big data sets of­ten seen as clunky oth­er­wise is very help­ful.

“The gov­ern­ment set aside around Rs 3,500 crore last year (be­fore re­vi­sion) for rur­al san­it­a­tion. The data on the spends and achieve­ments is avail­able on their web­site. Our ef­fort is to help trans­late these num­bers in­to a more un­der­stand­able form­at and we see the util­ity of hav­ing visu­al­isa­tions for this. This will help both de­cision makers and civil so­ci­ety un­der­stand pro­gress of rur­al san­it­a­tion at a glance. We hope that these can be then used as a gov­ernance tool to help de­cipher pat­terns, trends, good ex­amples and an­om­alies,” says Rangan.

Source: http://www.indianexpress.com/news/numbers-with-a-view/1098696/0

Leave a Reply