I cannot get xts to recognize Time Series in rownames

Question

I am having a problem with the xts package. I am trying to create an xts from a dataframe. For simplicity sake, I tried to replicate what I am trying to do on a small df below:

> df <- japanTOPIX[,1, drop = FALSE]
> typeof(df[,1])
[1] "double"
> typeof(rownames(df))
[1] "character"
> head(df, 3)
           X7164.JT.Equity
12/27/2000            65.0
12/28/2000            66.5
12/29/2000            66.2
> head(as.Date(rownames(df), format = "%m/%d/%Y"),3)
[1] "2000-12-27" "2000-12-28" "2000-12-29"
> timeBased(as.Date(rownames(df), format = "%m/%d/%Y"))
[1] TRUE
> xts(df, by = as.Date(rownames(df), format = "%m/%d/%Y"))
Error in xts(df, by = as.Date(rownames(df), format = "%m/%d/%Y")) : 
  order.by requires an appropriate time-based object
> head(strptime(rownames(df), format = "%m/%d/%Y"),3)
[1] "2000-12-27 EST" "2000-12-28 EST" "2000-12-29 EST"
> timeBased(strptime(rownames(df), format = "%m/%d/%Y"))
[1] TRUE
> xts(df, by = strptime(rownames(df), format = "%m/%d/%Y"))
Error in xts(df, by = strptime(rownames(df), format = "%m/%d/%Y")) : 
  order.by requires an appropriate time-based object
> head(as.POSIXlt(rownames(df), format = "%m/%d/%Y"),3)
[1] "2000-12-27 EST" "2000-12-28 EST" "2000-12-29 EST"
> timeBased(as.POSIXlt(rownames(df), format = "%m/%d/%Y"))
[1] TRUE
> xts(df, by = as.POSIXlt(rownames(df), format = "%m/%d/%Y"))
Error in xts(df, by = as.POSIXlt(rownames(df), format = "%m/%d/%Y")) : 
  order.by requires an appropriate time-based object
> head(as.POSIXct(rownames(df), format = "%m/%d/%Y"),3)
[1] "2000-12-27 EST" "2000-12-28 EST" "2000-12-29 EST"
> timeBased(as.POSIXct(rownames(df), format = "%m/%d/%Y"))
[1] TRUE
> xts(df, by = as.POSIXct(rownames(df), format = "%m/%d/%Y"))
Error in xts(df, by = as.POSIXct(rownames(df), format = "%m/%d/%Y")) : 
  order.by requires an appropriate time-based object

As you can see, I am getting an error in every attempt of trying to create this xts. The package claims I am not ordering by a time based object, but the 'by = ' variable is, in fact, time based. Any help?? (Please let me know if the example or code is not clear, and I will try to clarify.)

EDIT: I thought this may have to do with some non-unique values in the rownames, but that is not the case:

> nrow(df)
[1] 5115
> length(unique(rownames(df)))
[1] 5115

@BondedDust Hi Bonded. His data is of the form `12/27/2000 ` as you can see in `head(df, 3)`. Using `-` will produce ``s if I am not mistaken. — LyzandeR, Apr 23 '15 at 21:42

score 2 · Accepted Answer · answered Apr 23 '15 at 21:36

2

You are using the xts function incorrectly. The second argument is supposed to be order.by and not by. See the following example with your data:

Data:

df <- read.table(text='
           X7164.JT.Equity
12/27/2000            65.0
12/28/2000            66.5
12/29/2000            66.2')

Solution:

> xts(df, order.by = as.Date(rownames(df), format = "%m/%d/%Y"))
           X7164.JT.Equity
2000-12-27            65.0
2000-12-28            66.5
2000-12-29            66.2

And as you can see it works as intended!

P.S. instead of as.Date you can also use as.POSIXct or as.POSIXlt or strptime. All of them work correctly!

answered Apr 23 '15 at 21:36

LyzandeR

37,047
12
77
87

Wow. Every time. I am an idiot. Thanks for pointing this out! – lukehawk Apr 23 '15 at 22:25
Don't worry about it. It has happened to all of us plenty of times :P. We all need some help from time to time debugging. Glad I could be of help :) – LyzandeR Apr 23 '15 at 22:28

I cannot get xts to recognize Time Series in rownames

1 Answers1