ABSTRACT
When the Ministry of Civil Aviation (MoCA), Government of India, was launching its Ude Desh ka Aam Naagrik (UDAN) scheme, it wanted to know where new airports should be built. The UDAN scheme required airlines to bid for routes connecting airports that were unserved or underserved. While the industry did its own demand estimation, MoCA wanted to understand where it could expect demand. MoCA took a big data approach, working closely with a team at Google.
The project identified clusters where there was a willingness to travel coupled with the ability to pay. Google dove deep into its databases of searches and locations, meshing them with machine learning algorithms and artificial intelligence (yes, in 2017) to define cities/clusters where the data suggested there could be viable demand for an airport. However, an airport needs to be seen in context of the entire aviation network, requiring further parsing of data to balance the conflicting objectives of profitability of airports and accessibility for citizens concurrently.
How well did this work out? Of the ten cities identified (which were not shared with the industry), seven were picked by the industry in the first round of UDAN bidding as places from which they wanted to start flights. In the next round of UDAN bidding, 20 of the 30 cities identified were picked by the industry.
