SociocaRtography R · Machine Learning · GIS

Civic Tech: Policy formulation through Data Mining

Public data on cities is challenging to work on. Usually such data is collected and compiled by multiple government agencies as per their respective administrative boundaries. These boundaries have evolved in silos, are disparate and do not confirm to a homogeneous pattern of classification. For instance boundaries of Revenue, Police, Municipal, Postal, Electoral, Taxation units are very different from each other. The data from one source if supplemented with other can greatly enhance analytics and make decision making more accurate, efficient and effective. But disparate boundaries of these government agencies limits the potential of harnessing public information synergistically.

Urban-Economics

Fusing interdisciplinary data from government agencies, satellite imagery/nightlights and market data can mold innovative solutions. Emergence of megacities is characterized with rampant migration and capacity constraints. Rich, dynamic Big data based solutions on micro extracted land use features from satellite imagery holds promising insights for urban policy planning and governance. My work in the Indian Economic Survey, 2016-17 features the use of satellite imagery to improve cities revenue finance.

  1. LANDSAT Built-up density for improving cities finance revenue: Economic Survey of India The primary source of revenue for financing a city’s infrastructure and services comes from property tax collected by urban local bodies (ULB’s). But due to inaccurate enumeration, under-valuation and lack of adequate ground staff to monitor real-estate, there is severe under-collection of property tax across Indian cities. Using satellite imagery from LANDSAT program from joint National Aeronautics and Space Administration (NASA) and United States Geological Survey (USGS), I had worked with the office of the Chief Economic Adviser of India to estimate property tax potential for Bangalore and Jaipur. The satellite based raw data is geo-processed to identify built-up area, including everything from an independent housing unit to apartments as well as urban slums. Since the built-up wavelength bandwidth picks associated noise from road surface reflectance, a net built-up density measure was constructed to correct it. The building density on the ground provides an estimate of total build-up area (in square feet/km), which when interacted with zone specific guidance value of property tax per unit area gives an aggregate sum of potential property tax to be collected. The big data driven techno-policy solution has witnessed a welcome policy impact with expeditious adoption by municipal bodies in the intended spirit of competitive sub-federalism.

City & Crime

Crime has been an emerging feature of sprawling and abrupt urbanisation. Lumpen investment drive wealth concentration in cities that are racially and ethnically diverse. In-migration of service industry to this mix ferments towards a tiered social-geography. Crime in many case is an eventuality of frictional dialectics between these tiers.

  1. Delhi Crime: Voronoi Maps Criminal activities particularly in Urban regions are affected by location. The occurrence of a crime requires the juxtaposition of motivated offenders and suitable targets, a situation constrained in time and space.Indeed, “a limited number of sites, times, and situations constitute the space-time loci for the vast majority of offenses”(Brantingham, 2013) Spatially, Delhi is divided by series of intersecting/overlapping boundaries. There are over 7 districts, 21 towns, 400+ wards overlapped with over 200+ localities, 800+ sub-localities. In absense of centralized boundary, the shown polygons have been created from the location of the 280 police stations using Voronoi boundaries The Following maps showcases Delhi's reported crime incidents for 2014. In an attempt to create more awareness and aid police personnel deployment. Map Link


  2. Delhi Crime Maps: Micro Analysis @1sq. km Interactive map on Crime, Population, Night-light density, Total Crimes, Police Personnel depployed and actually sanctioned. The solution architecture is culminated from integrating micro satellite data from various open sources with event based data. Population from Socioeconomic Data & Applications Center (SEDAC, NASA) reporting population @ 1sq.km. Night-light radiance value collected from National Oceanic & Atmospheric Administration (NOAA). Police personnel deployed and reported crime incidents by category from United Residents Joint Action (URJA, NGO). Co-ordinates of the police stations/outposts were collected by geocoding each and every police station address from various API sources (Google, Bing etc). The said raw features are then geo-processed to assess Total criminal incidents, Population, Deployed and Sanctioned police force and Nightlights Radiance values @ 1sq.km. resolution for 2014. Map Link


  3. India Crime: Time Series Mapping Interactive application to study the trend of total reported Indian Penal Code(IPC) crime in India for 12 years. The data is collected from National Crime Recording Bearau for 626 districts. Map Link

I have worked on developing a semi-automated solution on similar lines, 'GRID'. By combining public census information with satellite imagery and commercial points of interest, a micro-geography intelligence model was developed estimating multiple socio-economic and demographic features. Data from overlapping boundaries was disaggregated in a spatially sensitized manner, to develop a repository for interdisciplinary analysis. Following are some of sample studies to aid Urban Planning. The next big idea is to therefore pool together from various sources, 'City Data Warehouse Project'.

Green/Building Spaces
Under Way

Post 2.