The Fifth Elephant 2013

An Event on Big Data and Cloud Computing

7 Ways to call elections using data

Submitted by Karthik Shashidhar (@karthiks) on Wednesday, 24 April 2013

Section: Analytics and Visualization Technical level: Beginner


Psephology has been turning more into a science than an art. This session explains how you can call elections using only data, even if you don't have much domain knowledge.


The recent presidential elections in the United States saw a landmark in that analyst Nate Silver managed to "call" (i.e. predict) the results in each of the 50 states accurately. This he did not using traditional psephology methods and "domain knowledge", but by simply using data and his superior number-crunching prowess.

This session will equip you with some quantitative tools that you can use to call elections purely based on data. Of course, you need to remember that the devil is in the implementation of these methods.



Speaker bio

Karthik Shashidhar is a freelance management consultant and data scientist. He is "resident quant" at Takshashila Institution, a public policy think tank, and teaches analytical methods and economics for Takshashila's Graduate Certificate in Public Policy course.

For the last two months he has been working on calling the forthcoming Karnataka Elections.



  • Govind Kanshi (@govindsk) 6 years ago

    Thanks Karthik - so we are going to get a treat of Bayesian world. So many “contexts” to be “formalized” into a “factor” - can’t wait to see this. So what is the prediction - # of seats vs particular people vs spread of votes....can’t wait that long.

Login with Twitter or Google to leave a comment