Converting ts object to data.frame - r

Question

I want to transform my ts object to data.frame object. My MWE is given below:
Code
set.seed(12345)
dat <- ts(data=runif(n=10, min=50, max=100), frequency = 4, start = c(1959, 2))
library(reshape2)
df <- data.frame(date=as.Date(index(dat)), Y = melt(dat)$value)
Output
date Y
1 1975-05-14 86.04519
2 1975-05-14 93.78866
3 1975-05-14 88.04912
4 1975-05-15 94.30623
5 1975-05-15 72.82405
6 1975-05-15 58.31859
7 1975-05-15 66.25477
8 1975-05-16 75.46122
9 1975-05-16 86.38526
10 1975-05-16 99.48685
I have lost my quarters in date columns. How can I figure out the problem?

Answer 1

How about
data.frame(Y=as.matrix(dat), date=time(dat))
This returns
Y date
1 86.04519 1959.25
2 93.78866 1959.50
3 88.04912 1959.75
4 94.30623 1960.00
5 72.82405 1960.25
6 58.31859 1960.50
7 66.25477 1960.75
8 75.46122 1961.00
9 86.38526 1961.25
10 99.48685 1961.50

Answer 2

yearmon (from zoo) allows creating Date objects.
> dat <- ts(data=runif(n=10, min=50, max=100), frequency = 4, start = c(1959, 2))
> data.frame(Y=as.matrix(dat), date=as.Date(as.yearmon(time(dat))))
Y date
1 51.72677 1959-04-01
2 57.61867 1959-07-01
3 86.78425 1959-10-01
4 50.05683 1960-01-01
5 69.56017 1960-04-01
6 73.12473 1960-07-01
7 69.40720 1960-10-01
8 70.12426 1961-01-01
9 58.94818 1961-04-01
10 97.58294 1961-07-01

Answer 3

The package timetk has several conversion functions. In your case:
dat <- ts(data=runif(n=10, min=50, max=100), frequency = 4, start = c(1959, 2))
timetk::tk_tbl(dat)
# A tibble: 10 x 2
index value
<S3: yearqtr> <dbl>
1 1959 Q2 86.04519
2 1959 Q3 93.78866
3 1959 Q4 88.04912
4 1960 Q1 94.30623
5 1960 Q2 72.82405
6 1960 Q3 58.31859
7 1960 Q4 66.25477
8 1961 Q1 75.46122
9 1961 Q2 86.38526
10 1961 Q3 99.48685

Answer 4

Seems that converting from xts objects seems to be both reliable and well documented. Below works and with the new date column in date / yearqtr class.
library(xts)
datx <- as.xts(dat)
df <- data.frame(date=index(datx), coredata(datx))
Checking class of date:
class(df$date)
[1] "yearqtr"
And result:
print(df)
date coredata.datx.
1 1959 Q2 86.04519
2 1959 Q3 93.78866
3 1959 Q4 88.04912
4 1960 Q1 94.30623
5 1960 Q2 72.82405
6 1960 Q3 58.31859
7 1960 Q4 66.25477
8 1961 Q1 75.46122
9 1961 Q2 86.38526
10 1961 Q3 99.48685

Answer 5

Package 'ggpp' provides function try_data_frame() (implemented using packages 'xts', 'zoo' and 'lubridate') that does the conversion in a single step. (This function is used in package 'ggpp' to implement a ggplot() method for time series, and returns the time index converted into a class that packages 'ggplot2' and 'scales' can use: Date or POSIXct.)
set.seed(12345)
dat.ts <- ts(data=runif(n=10, min=50, max=100), frequency = 4, start = c(1959, 2))
library(ggpp)
#> Loading required package: ggplot2
#>
#> Attaching package: 'ggpp'
#> The following object is masked from 'package:ggplot2':
#>
#> annotate
dat.df <- try_data_frame(dat.ts)
str(dat.df)
#> 'data.frame': 10 obs. of 2 variables:
#> $ time : Date, format: "1959-05-01" "1959-08-01" ...
#> $ dat.ts: num 86 93.8 88 94.3 72.8 ...
dat.df
#> time dat.ts
#> 1 1959-05-01 86.04519
#> 2 1959-08-01 93.78866
#> 3 1959-11-01 88.04912
#> 4 1960-02-01 94.30623
#> 5 1960-05-01 72.82405
#> 6 1960-08-01 58.31859
#> 7 1960-11-01 66.25477
#> 8 1961-02-01 75.46122
#> 9 1961-05-01 86.38526
#> 10 1961-08-01 99.48685
Created on 2022-09-03 with reprex v2.0.2
See help(try_data_frame()) for the details on how to set the names of columns or alter the way in which dates or times are handled.