compute rate by group - r

Question

I have a data that want to calculate the growth rate by the previous year and quarter.
# dt
yq A B
2013 Q1 35233684 270950851
2013 Q2 36235895 274194641
2013 Q3 36767497 275614372
2013 Q4 37273346 277125049
2014 Q1 37788578 278202677
2014 Q2 38674955 281025545
str(dt)
Classes ‘data.table’ and 'data.frame': 6 obs. of 3 variables:
$ yq : 'yearqtr' num 2013 Q1 2013 Q2 2013 Q3 2013 Q4 ...
$ A : int 35233684 36235895 36767497 37273346 37788578 38674955
$ B: int 270950851 274194641 275614372 277125049 278202677 281025545
- attr(*, ".internal.selfref")=<externalptr>
The code I use:
dt[, lapply(.SD, function(x)x/shift(x) - 1), .SDcols = 2:3, by = .(quarter(yq))]
quarter A B
1 NA NA
1 0.07251283 0.02676436
2 NA NA
2 0.06731060 0.02491261
3 NA NA
4 NA NA
I got the result; however, I want the format like this:
I want it to keep the column yq and order with year and quarter.
yq A B
2013 Q1 35233684 270950851
2013 Q2 36235895 274194641
2013 Q3 36767497 275614372
2013 Q4 37273346 277125049
2014 Q1 37788578 278202677
2014 Q2 38674955 281025545
yq A B A_R B_R
2013 Q1 35233684 270950851 NA NA
2013 Q2 36235895 274194641 NA NA
2013 Q3 36767497 275614372 NA NA
2013 Q4 37273346 277125049 NA NA
2014 Q1 37788578 278202677 0.07251283 0.02676436
2014 Q2 38674955 281025545 0.06731060 0.02491261
How do I do to edit my code?
# Data
library(data.table)
dt <- fread("yq A B
2013 Q1 35233684 270950851
2013 Q2 36235895 274194641
2013 Q3 36767497 275614372
2013 Q4 37273346 277125049
2014 Q1 37788578 278202677
2014 Q2 38674955 28102554", header = T)

Answer 1

So I see you are using the zoo package and the function yearqtr. I am unable to get the yq column read using your fread but I just quickly reproduced the data as follows:
library(zoo)
dt<-data.table(cbind(yq=2013 + seq(0,5)/4,
A = c(35233684, 36235895, 36767497, 37273346, 37788578, 38674955),
B = c(270950851, 274194641, 275614372, 277125049, 278202677, 281025545)))
Then just converted the yq as follows:
dt[,yq:=as.yearqtr(yq)]
Now if you want to keep that column you will need to update the columns by specifying them:
cols<-c("A","B")
dt[,eval(cols):=lapply(.SD,function(x)x/shift(x) - 1), .SDcols = 2:3, by = .(quarter(yq))]
So simply add as many columns as you need to the cols vector and use eval so data.table will not create a new column named "cols"! Does this answer your question?

Answer 2

I am not familiar with data.table package. But here is how I would do it using dplyr.
You can first separate your yq column into two columns, y and q. I skipped this step in my code because I don't know what exact datatype you used in the original data.
Then group by q to do the calculation.
library(data.table)
dt <- fread(
"y q A B
2013 Q1 35233684 270950851
2013 Q2 36235895 274194641
2013 Q3 36767497 275614372
2013 Q4 37273346 277125049
2014 Q1 37788578 278202677
2014 Q2 38674955 281025545", header = T)
library(tidyverse)
dt%>%group_by(q)%>%
arrange(y)%>%
mutate(growth_rate_over_year_A= A/lag(A)-1,
growth_rate_over_year_B= B/lag(B)-1)%>%
ungroup
output:
# A tibble: 6 x 6
y q A B growth_rate_over_year_A growth_rate_over_year_B
<int> <chr> <int> <int> <dbl> <dbl>
1 2013 Q1 35233684 270950851 NA NA
2 2013 Q2 36235895 274194641 NA NA
3 2013 Q3 36767497 275614372 NA NA
4 2013 Q4 37273346 277125049 NA NA
5 2014 Q1 37788578 278202677 0.0725 0.0268
6 2014 Q2 38674955 281025545 0.0673 0.0249