Home/ Questions/Q 7930521
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-03T20:27:15+00:00

I regularly build a data frame with time values (originating from xts or zoo

  • 0

I regularly build a data frame with time values (originating from xts or zoo objects) for analysis and plotting. Here’s an example:

library(xts)   

x1 <- xts(rnorm(10), as.Date("2012-01-01") + 0:9)   
x2 <- xts(rnorm(10), as.Date("2012-01-04") + 0:9)   

(df <- data.frame(merge(x1=x1, x2=x2), v1=4:16, v2=rnorm(13)))    

                   x1          x2 v1          v2   
2012-01-01  0.1930827          NA  4  1.05972724   
2012-01-02  0.4429592          NA  5 -1.89299068   
2012-01-03  1.6657630          NA  6  0.70445966  
2012-01-04 -0.2765922 -0.26728223  7  0.35336959   
2012-01-05 -0.1756590 -2.04888130  8  0.90129924  
2012-01-06 -2.4849132  0.49400975  9  1.54486914  
2012-01-07 -0.9993353 -1.09308203 10  1.16600015  
2012-01-08 -0.7326309  0.55781566 11  0.37178542  
2012-01-09 -0.2973543 -0.59872496 12  0.07512468  
2012-01-10 -1.5061380  0.08567125 13  1.77494367  
2012-01-11         NA  0.81835375 14 -0.38211167  
2012-01-12         NA  1.30131894 15 -1.09220795  
2012-01-13         NA -1.29505649 16  1.27148069

To plot the time series, I need to use as.Date(row.names(df)) for the horizontal axis. For example:

plot(as.Date(row.names(df)), df$x1, type="l", xlab="", ylab="Test", main="Using row.names()", col="red")   
lines(as.Date(row.names(df)), df$x2, col="blue")    
lines(as.Date(row.names(df)), df$v2, col="green")

It’s a real pain to use as.Date(row.names(df)) over and over, so in the past, I added a column to the data frame using the following code:

(df <- cbind(df, tim=as.Date(row.names(df))))

I can now plot the time series as:

with(df, {  
  plot(tim, x1, type="l", xlab="", ylab="Test", main="Using a new tim column", col="red")  
  lines(tim, x2, col="blue")  
  lines(tim, v2, col="green")  
})

This technique works, but adding that tim column feels sloppy. Is there a better/cleaner way to do this?

Edit 1 (2012-05-13) ====================================

The reason I’m using a data frame is because this data is commonly fed to one or more packages that require a data frame, and/or because I typically need to represent non-numeric data in some of the variables.

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    I would stick with the zoo facilities as much as possible:

    library(xts)   
x1 <- xts(rnorm(10), as.Date("2012-01-01") + 0:9)   
x2 <- xts(rnorm(10), as.Date("2012-01-04") + 0:9)   

dz <- merge(x1=x1, x2=x2); dz$v1=4:16; dz$v2=rnorm(13)    
 # keeps it a zoo object rather than a data.frame

with(dz, plot( x1, type="l", xlab="", ylab="Test", main="Not using row.names()\nStick with zoo" ) )
lines( dz$x2, col="blue")    
lines( dz$v2, col="green")

    enter image description here

    • 0