What's in a name? Inferring religious community from South Asian names

Religion in South Asia is fascinating, but hard to quantify. India for example is the third largest Muslim country worldwide. But where do Muslims live, what do they earn, how do they vote? Unfortunately, official data can't tell - they're too coarse.Luckily, name lists are more readily available, for example in electoral registers. We know: a voter called "Ahmad" is quite likely a Muslim, a "Bhavna" most likely not. Software can accurately model such connotations; across millions of voters, errors cancel each other out. I thus estimated the share of voters with "Muslim" names in 893.359 polling booths, replacing official data for 676 districts: much more detail!