Question

I don’t understand why I can’t find a solution for this, since I feel that this is a pretty basic question. Need to ask for help, then. I want to rearrange airquality dataset by month with maximum temp value for each month. In addition I want to find the corresponding day for each monthly maximum temperature. What is the laziest (code-wise) way to do this?

I have tried following without a success:

require(reshape2)
names(airquality) <- tolower(names(airquality))
mm <- melt(airquality, id.vars = c("month", "day"), meas = c("temp"))

dcast(mm, month + day ~ variable, max)
aggregate(formula = temp ~ month + day, data = airquality, FUN = max)

I am after something like this:

month day temp
5     7    89
...

Answer 1

There was quite a discussion a while back about whether being lazy is good or not. Anwyay, this is short and natural to write and read (and is fast for large data so you don’t need to change or optimize it later) :

require(data.table)
DT=as.data.table(airquality)

DT[,.SD[which.max(Temp)],by=Month]

     Month Ozone Solar.R Wind Temp Day
[1,]     5    45     252 14.9   81  29
[2,]     6    NA     259 10.9   93  11
[3,]     7    97     267  6.3   92   8
[4,]     8    76     203  9.7   97  28
[5,]     9    73     183  2.8   93   3

.SD is the subset of the data for each group, and you just want the row from it with the largest Temp, iiuc. If you need the row number then that can be added.

Or to get all the rows where the max is tied :

DT[,.SD[Temp==max(Temp)],by=Month]

     Month Ozone Solar.R Wind Temp Day
[1,]     5    45     252 14.9   81  29
[2,]     6    NA     259 10.9   93  11
[3,]     7    97     267  6.3   92   8
[4,]     7    97     272  5.7   92   9
[5,]     8    76     203  9.7   97  28
[6,]     9    73     183  2.8   93   3
[7,]     9    91     189  4.6   93   4