Population data was gathered from the Indian decennial census. Because the last comprehensive census in India occurred in 2011, we pulled population data from 1951-2011 in ten-year intervals to build a projection model to create population projections.

We used two models for these population projections: The Monte Carlo and the Autoregression Integrated Moving Average (ARIMA) models. In each of India’s metro areas, we tried to identify the best method depending on the type of growth seen. In both approaches, we use the 1951-2001 data to build models that would predict 2011. The model with the most accurate prediction in each area won, and we used that model to project further population figures.

India’s metropolitan areas were hand-curated based on the country’s districts. Note that there have been significant definitional changes over the time series that we gathered and thus we had to adapt our projections to reflect the most accurate geographical definition at each time period.

India’s occupation data was extracted from the country’s 2011 census conducted by the Office of the Registrar General & Census Commissioner of India. This census gave us occupational data for a total of 630 geographical areas. India uses its own occupational classification system NCO-2004 which is similar to the International Standard Classification of Occupations (ISCO-88). Lightcast converted this classification to ISCO-88 before projecting the data.

Finally, to compile supply projections, we took the expected population growth rate from the models mentioned above and applied that same growth rate across the LMI in each of India’s districts. Once we had the projections, we mapped all the occupational categories we had translated to International Standard Classification of Occupations (ISCO-08) to Global Occupations.