Some data gymnastics using data from http://covid19india.org 

Calculated doubling rate of the number of cases using 3-day growth rates (to smooth the curve). Interesting that growth has slowed incredibly since we went on lockdown

The week leading up to the "junta curfew" saw an acceleration in cases, from doubling once in 7 days to doubling once in less than 3 days (I use continuously compounded growth rates).

After that things have slowed signfiicantly

Calculated growth rate as ln(Cases/lag(Cases,3)), and the doubling days as ln(2)/GrowthRate*3
(the last * 3 required to normalise the rate to "per day").

Today's data point is incomplete and hence an anamoly, but we seem to be on the right track. Hopefully thigns will get under control before the lockdown ends.

Re ran the numbers (figured out how to use the API so no more copy-pasting! Thanks guys!).

Seems like now we're at a sort of stable state of doubling every 4.5 days or so

Ran it once again this morning. Seems like the "improvement" to doubling only once in 4.5 days was a false dawn. Discovery of the Nizamuddin cluster means we're back to ~3 day doubling again. Should continue until all those cases are flushed out

And this is not necessarily a bad thing - remember that the Nizamuddin conference ended around March 20th. So we're discovering infections that actually happened a long time ago.

My worry now is there are more such possible clusters which haven't surfaced yet.

Thing that still gives me hope from the @covid19indiaorg data - most cases seem to have stemmed out of long and sustained exposure to infected people (family, roommates, colleagues, etc).

On @rahulrg 's request, decided to do this by state as well. Restricted it for states that have had >= 100 infections.

This, remember, is effective "doubling time". Notice how Kerala has "recovered"!

Update:

At the end of yesterday, the growth of cases in India had marginally slowed, to a 3.6 day doubling

At the state level, it looks like Kerala is in the process of bringing it under control. Doubling in more than 10 days now

It's also interesting to see the "notes" on the cases as per @covid19indiaorg . Essentially it seems like most new cases are still:
1. People who returned from abroad before the airports were shut
2. Nizamuddin cluster
3. Household members of people already infected

And that gives hope that Nizamuddin apart, the virus hasn't been all that viral in India. Hopefully there are no other such clusters that we don't yet know about.

I'm making this a sort of daily update now

We seem to have turned a corner again in doubling. After a few days of doubling days in the threes, we're back to four.

Kerala is surely getting things under control now (unless, of course, another bunch like Nizamuddin is discovered). Karnataka getting there as well.

I've also started tracking cases where "Delhi" appears in notes. Most of this relates to the Nizamuddin cluster.

Some good news - we seem to have turned the corner. Yesterday's growth in number of cases implies a doubling in about 4.8 days. Up from ~4 days for last 4-5 days

And state-wise:

Kerala should lift the lockdown on the 15th. Maybe Karnataka as well, to a limited extent. TN also seems fine apart from the TJ cases.

Update on morning of 8/4:

The good news continues. Doubling time now up to 5.6 days. Back to the sort of trend we had before the Nizamuddin cases started being unearthed

State wise, Delhi has joined Kerala and Karnataka in turning a corner. Mumbai also seems to have improved. Need to see if it will sustain. Mumbai is going to be crucial because of its insanely high density.

Good morning.

The numbers continue to improve. Based on the latest data, India cases are doubling every 6.5 days now. Also very few fresh cases related to the Nizamuddin thing

And Southern states recovery continues as well. The new points of concern are Haryana and Punjab, where cases have been growing rapidly over the last few days.

The bad news continues. Doubling time continues to go down secularly for 3 days now. Down to 5.8 now.

And there seems to be a massive state-wise divergence now. Southern states (except TN) are all holding up really well. Haryana also suddenly improved!

Rajasthan and Gujarat a big matter of concern now.

Hopefully this divergence means that the decision to extend lockdown will get decentralised to the state (or even district) level. One size fits all clearly doesn't work