R Notebook

Data

I’m getting data that was put together on Kaggle. There must be a better source, no?

	554	588	823	707	260
Sno	554	588	823	707	260
Date	2020-04-01	2020-04-03	2020-04-10	2020-04-06	2020-03-21
Time	7:30 PM	6:00 PM	5:00 PM	6:00 PM	6:00 PM
State.UnionTerritory	Tamil Nadu	Andhra Pradesh	Odisha	Uttar Pradesh	Puducherry
ConfirmedIndianNational	-	-	-	-	1
ConfirmedForeignNational	-	-	-	-	0
Cured	6	1	2	21	0
Deaths	1	1	1	3	0
Confirmed	234	132	44	305	1
State	Tamil Nadu	Andhra Pradesh	Odisha	Uttar Pradesh	Puducherry

The data set has daily observations. However, due to reporting issues and glitches daily level data can have substantial noise. Best to group things up at the weekly level to smooth this out a bit.

State	Chandigarh	Tamil Nadu	Puducherry	Maharashtra	Tamil Nadu
Week	2020-04-12	2020-04-12	2020-03-22	2020-03-22	2020-03-15
Confirmed.week	145	8253	7	792	12
Deaths.week	0	88	0	21	0

Trends across State and Time

Let’s see how many confirmed cases.

The same with Deaths per week.

Besides the pure numbers of daily or weekly new cases and deaths, it is also useful to consider the percentage of deaths (relative to known or confirmed infections). To compute this rate we need to recognize that deaths lag infection by some time period. Here we’ll consider a 2-week lag. And now a look at the death rate and how it is changing over time.

Taking a different view let’s plot a line chart across time, but limit the number of states in it to only those states with a high number of total infections (cumulative cases > 500,000)

And here’s the death rate among the states with fewer infections.