SociocaRtography R · Machine Learning · GIS

Parth Khare

DRW Fellow, Clinton-Global Fellow, MSCAPP:University of Chicago

Parth Khare

I am an economist and a data scientist, with 7+ years of experience in evidence based policy research and big-data analytics across Multilateral Development Bank, Government of India’s apex economic research wing, Think Tanks, and MNCs. I have formerly worked with the Chief Economic Adviser of India, advocating Big-Data & GIS driven economic policy in the Government of India. My interest & work is largely focused at the intersection of high frequency satellite data & heavy volumes of census, biometric & retail data.

My core work is focused on innovating socio-demographic indicators for countries where conventional census data is either lagged or not available at micro resolution. To this end, I have published following data-policy, civic-tech solutions with the Indian Government, featured in the Economic Survey, which have been successfully implemented or in process of being implemented at national/sub-national levels. Long term impact of Climate Change on Indian Agriculture (with 60+ years of data & 10x number of weather stations than IPCC/GHCN repositories:VoxDev. GIS driven property tax collection by accurately identifying built-up units with subjectively tuned . Dynamic tracking of India's Urbanisation rate (Pg 221-) (using satellite based Built-up & Nightlights data).

I have conducted training sessions on R & Decision Trees to Indian government bureaucrats & in the private sector. Duing my term as Manager, Data Science at Nielsen, I had coded a non-parametric model architecture GRID in R, which estimates socio-economic attributes at 1 square km level precision. The algorithm, implemented on Asian & European geographies worked by interacting public census data with micro satellite imagery data to unearth spatially intelligent patterns. The hidden potential of layering multiple data sources had motivated me on developing open source solution to aid public policy. Some interactive map applications on Indian data: India Crime Analysis, Delhi Crime Deployment, R Shiny Maps & Voronoi Maps.

I have developed & customized geo-spatial functions in R and have automated a wrapper function, GBM optimizor for efficient parameters selection (trees, shrinkage, OOB). R Utility Kit is an R attempt to streamline codes on frequently used data operations.