Monthly operations time series with apply.monthly in R - r

Question

The problem is to use apply.monthly or any other similar function to do monthly operations with a dataset. The data I have looks like the following:
> minidata[1:10,]
date Month Year TMIN
1 1948-01-01 Jan 1948 1.1
2 1948-01-02 Jan 1948 7.2
3 1948-01-03 Jan 1948 5.0
4 1948-01-04 Jan 1948 9.4
5 1948-01-05 Jan 1948 4.4
> tail(minidata)
date Month Year TMIN
54 1948-02-23 Feb 1948 2.8
55 1948-02-24 Feb 1948 -0.6
56 1948-02-25 Feb 1948 1.7
57 1948-02-26 Feb 1948 2.8
58 1948-02-27 Feb 1948 4.4
59 1948-02-28 Feb 1948 3.3
Task, use my own function to produce the monthly mean:
mymean <- function(date){
for (j in 1:days_in_month(date)){
avg = (1/(days_in_month(date))
*sum(minidata$TMIN[1:days_in_month(date)])}
return(avg)
}
The result must be the same as the R function in the xts package:
dat.xts <- xts(x= minidata$TMIN,order.by = minidata$date)
> apply.monthly(dat.xts,mean)
[,1]
1948-01-31 2.312903
1948-02-28 2.082143
My function outputs the correct values:
> mymean(minidata$date[1])
Jan
2.312903
> mymean(dat.xts[1])
Jan
2.312903
I wouldn't mind if $apply.monthly$ generated a new column with the means, but I have to use my own function! (This is an example, in reality my function is a lot harder).
I tried:
> apply.monthly(dat.xts,function(dat.xts) mymean(dat.xts))
Error in coredata.xts(x) : currently unsupported data type
In addition: There were 50 or more warnings (use warnings() to see the first 50)
Thanks!
Update: days_in_month can be found in the lubridate package. It calculates the number of days in a given month

Answer 1

Your function is the issue, not apply.monthly. I don't know where the days_in_month function is defined, but it probably doesn't work with xts objects. I assume it expects a date-time class.
And your mymean function references an object that isn't passed to it, which is not good practice because it makes R search for minidata.
Your function should expect an xts object containing a month of data and operate only on that data, not some object outside the function scope. For example:
mymean <- function(Data) {
days <- days_in_month(index(Data)[1])
avg <- (1/days) * sum(Data$Close)
return(avg)
}
require(xts)
data(sample_matrix)
x <- as.xts(sample_matrix)
apply.monthly(x, mymean)

Answer 2

To perform operations within groups of a data frame, you can use the dplyr package. For instance, to get the average TMIN within each group:
library(dplyr)
summarize(group_by(minidata, Month), mean = mean(TMIN))
This is often written as:
minidata %>% group_by(Month) %>%
summarize(mean = mean(TMIN))

Answer 3

Your function works only on data frames, the xts object is different and isn't going to work the way you want. That is why it is giving you errors.
Besides that, you do not want to do this with a loop. It is going to take much longer than many other ways of doing it.
David's answer (use dplyr::group_by and dplyr::summarize) is the best way to handle this. You can use a custom function in the summarize if that is what the problem is. Just define your function and use it there.