converting a data frame (panel data) to zoo

Question

How can I convert the following a data frame (panel data) to zoo? let df indicates my data frame;

Ctry  year   Carx   Brx
 A    2000    23     12
 A    2001    18     16
 A    2002    20     20
 A    2003    NA     18
 A    2004    24     NA
 A    2005    18     12
 B    2000    NA     22
 B    2001    NA     20
 B    2002    NA     14
 B    2003    NA     NA
 B    2004    18     NA
 B    2005    16     14   
 C    2000    NA     NA
 C    2001    NA     25
 C    2002    24     32
 C    2003    21     NA
 C    2004    NA     15
 C    2005    24     NA

I have tried the following code

df.zoo=zoo(df[,-2], order.by=as.Date(strptime(as.character(df[,2]), "%Y")))

I obtain the following result with a warning: "some methods for “zoo” objects do not work if the index entries in ‘order.by’ are not unique"

    year     Ctry   Carx   Brx
 2000-06-02    A    23     12
 2000-06-02    B    NA     22
 2000-06-02    C    NA     NA
    ..        ..    ..     ..

What is the most suitable form of conversion for the above data frame to perform functions in zoo package?

many thanks for your help...

G. Grothendieck G. Grothendieck · Accepted Answer · 2014-06-02T10:33:06

zoo with split We cannot have zoo objects which look like the sample output in the question; however, if we change the question to require a wide form time series we can use read.zoo with the split argument:

z <- read.zoo(DF, split = "Ctry", index = "year", FUN = identity)

giving:

> z
     Carx.A Brx.A Carx.B Brx.B Carx.C Brx.C
2000     23    12     NA    22     NA    NA
2001     18    16     NA    20     NA    25
2002     20    20     NA    14     24    32
2003     NA    18     NA    NA     21    NA
2004     24    NA     18    NA     NA    15
2005     18    12     16    14     24    NA

two zoo objects Alternately, consider creating two zoo objects:

Carx.z <- read.zoo(DF[, -3], split = "Ctry", index = "year", FUN = identity)
Brx.z <- read.zoo(DF[, -4], split = "Ctry", index = "year", FUN = identity)

giving:

> Carx.z
      A  B  C
2000 12 22 NA
2001 16 20 25
2002 20 14 32
2003 18 NA NA
2004 NA NA 15
2005 12 14 NA
> Brx.z
      A  B  C
2000 23 NA NA
2001 18 NA NA
2002 20 NA 24
2003 NA NA 21
2004 24 18 NA
2005 18 16 24

Note: In the development version of zoo read.zoo is able to figure out the FUN = identity part automatically so that argument could be dropped.

Reproducible test data. We used the following data frame DF in the above:

Lines <- "Ctry  year   Carx   Brx
 A    2000    23     12
 A    2001    18     16
 A    2002    20     20
 A    2003    NA     18
 A    2004    24     NA
 A    2005    18     12
 B    2000    NA     22
 B    2001    NA     20
 B    2002    NA     14
 B    2003    NA     NA
 B    2004    18     NA
 B    2005    16     14   
 C    2000    NA     NA
 C    2001    NA     25
 C    2002    24     32
 C    2003    21     NA
 C    2004    NA     15
 C    2005    24     NA
"
DF <- read.table(text = Lines, header = TRUE)

converting a data frame (panel data) to zoo

1 Answers