Question

x is an xts object full of data; let’s assume OHLC data for the sake of example.
I want to create another xts object, with the same size and datestamps, but different columns (e.g. some indicators).

My current approach feels crude:

a = x$close
for(nn in 1:10){
    z = analysis(x,nn) #Returns an enhanced version of x
    z2 = z$result   #Get out just the data I want, so I can rename the column
    colnames(z2) = paste("result",nn,sep="_")
    a = cbind(a,z2) #Merge in each result
    }
a$close = NULL  #Tidyup

I.e. I bring in just one column from x, any old column, just to get the structure, then throw that away at the end. (It works, so I’m happy, but it feels like there must be a better way.)

I tried some ideas like this:

a = xts(index(x))
a = xts(orderby=index(x))
a = as.xts(index(x))
a = as.xts(orderby=index(x))

But they give me empty XTS objects. E.g. when I then try this I get an error:

a$dummy = 1

Answer 1

Note the argument is order.by, not orderby. That doesn’t solve your issue though. What you’re looking for is:

a <- xts(order.by=index(x))
a <- merge(a, dummy=1)

a$dummy <- 1 doesn’t work because zoo objects can be a vector or a matrix, while xts objects are always a matrix and there’s no $<-.xts method.