How to repeat observations forward?

Question

I want to repeat the values of one variable conditionally. For example, I have this data.frame

      cod  ano partido_prefeito
1  110001 1998             <NA>
2  110001 1999             <NA>
3  110001 2000             <NA>
4  110001 2001              PPB
5  110001 2002             <NA>
6  110001 2003             <NA>
7  110001 2004             <NA>
8  110001 2005              PDT
9  110001 2006             <NA>
10 110001 2007             <NA>
11 110001 2008             <NA>
12 110001 2009              PTN
13 110001 2010             <NA>
14 110001 2011             <NA>
15 110001 2012             <NA>
16 110001 2013             PMDB
17 110001 2014             <NA>
18 110001 2015             <NA>
19 110001 2016             <NA>
20 110002 1998             <NA>

For the variable "partido_prefeito" I have a lot of "NA" observations, however, I want to repeat the observation for the next 3 years, until the observation changes, for each "cod". Turning into something like that:

      cod  ano partido_prefeito
1  110001 1998             <NA>
2  110001 1999             <NA>
3  110001 2000             <NA>
4  110001 2001              PPB
5  110001 2002              PBP
6  110001 2003              PBP
7  110001 2004              PBP
8  110001 2005              PDT
9  110001 2006              PDT
10 110001 2007              PDT
11 110001 2008              PDT
12 110001 2009              PTN
13 110001 2010              PTN
14 110001 2011              PTN
15 110001 2012              PTN
16 110001 2013             PMDB
17 110001 2014             PMDB
18 110001 2015             PMBD
19 110001 2016             PMBD
20 110002 1998             <NA>

For the first 3 years: 1998, 1999, 2000, the data will still be "NA". The important detail is that I have a lot of observations with different "cod". How can I easily do this transformation?

Have a look at `zoo::na.locf()` – coffeinjunky Oct 10 '18 at 22:03 — coffeinjunky, Oct 10 '18 at 22:03

score 1 · Accepted Answer · answered Oct 10 '18 at 22:09

Perfect time to use fill from tidyverse.

First, make sure your <NA> values are actual NAs and not strings. Then:

library(tidyverse)

data %>% group_by(cod) %>% fill(partido_prefeito)

fill takes the last value and fills it down. The only requirement for this to work is that you have to have actual NA values in your data. If NA is stored as a string, i.e. "<NA>" you'll need to convert it to <NA> first.

 1 110001  1998 <NA>            
 2 110001  1999 <NA>            
 3 110001  2000 <NA>            
 4 110001  2001 PPB             
 5 110001  2002 PPB             
 6 110001  2003 PPB             
 7 110001  2004 PPB             
 8 110001  2005 PDT             
 9 110001  2006 PDT             
10 110001  2007 PDT             
11 110001  2008 PDT             
12 110001  2009 PTN             
13 110001  2010 PTN             
14 110001  2011 PTN             
15 110001  2012 PTN             
16 110001  2013 PMDB            
17 110001  2014 PMDB            
18 110001  2015 PMDB            
19 110001  2016 PMDB            
20 110002  1998 <NA>

Data:

data <- structure(list(cod = c(110001L, 110001L, 110001L, 110001L, 110001L, 
110001L, 110001L, 110001L, 110001L, 110001L, 110001L, 110001L, 
110001L, 110001L, 110001L, 110001L, 110001L, 110001L, 110001L, 
110002L), ano = c(1998L, 1999L, 2000L, 2001L, 2002L, 2003L, 2004L, 
2005L, 2006L, 2007L, 2008L, 2009L, 2010L, 2011L, 2012L, 2013L, 
2014L, 2015L, 2016L, 1998L), partido_prefeito = structure(c(NA, 
NA, NA, 3L, NA, NA, NA, 1L, NA, NA, NA, 4L, NA, NA, NA, 2L, NA, 
NA, NA, NA), .Label = c("PDT", "PMDB", "PPB", "PTN"), class = "factor")), class = "data.frame", row.names = c("1", 
"2", "3", "4", "5", "6", "7", "8", "9", "10", "11", "12", "13", 
"14", "15", "16", "17", "18", "19", "20"))

How to repeat observations forward?

1 Answers1

Data: