Fill numeric variable while preserving group - r

Question

[EDITED to reflect a better example]
Say I have a dataframe like this:
df <- data.frame(x = c("A","A","B", "B"), year = c(2001,2004,2002,2005))
> df
x year
1 A 2001
2 A 2004
3 B 2002
4 B 2005
How can I increment year by 1 while preserving x? I would like to fill in year so that the sequence is this:
x year
1 A 2001
2 A 2002
3 A 2003
4 A 2004
5 B 2002
6 B 2003
7 B 2004
8 B 2005
Can anyone recommend a good way of doing this?
#useR recommend this approach:
> data.frame(year = min(df$year):max(df$year)) %>%
full_join(df) %>%
fill(x)
Joining, by = "year"
year x
1 2001 A
2 2002 B
3 2003 B
4 2004 A
5 2005 B
However that does not match the desired output.

Answer 1

An option using tidyr::complete and dplyr::lead can be as:
library(tidyverse)
df <- data.frame(x = LETTERS[1:3], year = c(2001,2004,2007))
df %>% mutate(nextYear = ifelse(is.na(lead(year)),year, lead(year)-1)) %>%
group_by(x) %>%
complete(year = seq(year, nextYear, by=1)) %>%
select(-nextYear) %>%
as.data.frame()
# x year
# 1 A 2001
# 2 A 2002
# 3 A 2003
# 4 B 2004
# 5 B 2005
# 6 B 2006
# 7 C 2007
Edited: The solution for modified data
df <- data.frame(x = c("A","A","B", "B"), year = c(2001,2004,2002,2005))
library(tidyverse)
df %>% group_by(x) %>%
complete(year = seq(min(year), max(year), by=1)) %>%
as.data.frame()
# x year
# 1 A 2001
# 2 A 2002
# 3 A 2003
# 4 A 2004
# 5 B 2002
# 6 B 2003
# 7 B 2004
# 8 B 2005

Answer 2

Using base R (with a little help from zoo):
full_df = data.frame(year = min(df$year):max(df$year))
df = merge(df, full_df, all = TRUE)
df = df[order(df$year), ]
df$x = zoo::na.locf(df$x)
df
# year x
# 1 2001 A
# 2 2002 A
# 3 2003 A
# 4 2004 B
# 5 2005 B
# 6 2006 B
# 7 2007 C
Using the "tidyverse"
df <- data.frame(x = LETTERS[1:3], year = c(2001,2004,2007))
library(dplyr)
library(tidyr)
df = df %>% mutate(year = factor(year, levels = min(year):max(year))) %>%
complete(year) %>%
fill(x) %>%
mutate(year = as.numeric(as.character(year)))
df
# # A tibble: 7 x 2
# year x
# <dbl> <fctr>
# 1 2001 A
# 2 2002 A
# 3 2003 A
# 4 2004 B
# 5 2005 B
# 6 2006 B
# 7 2007 C

Answer 3

We can first split by x, then create a year vector for each x group, join with each group df, fill down x, then finally rbind all group df's together.
library(dplyr)
library(tidyr)
df %>%
split(.$x) %>%
lapply(function(y) data.frame(year = min(y$year):max(y$year)) %>%
full_join(y) %>%
fill(x)) %>%
unname() %>%
do.call(rbind, .)
Result:
year x
1 2001 A
2 2002 A
3 2003 A
4 2004 A
5 2002 B
6 2003 B
7 2004 B
8 2005 B

Answer 4

Here's a pretty simple base R method with tapply and stack.
stack(tapply(df$year, df["x"], function(x) min(x):max(x)))
Here, tapply splits the year vector by df$x groups and then constructs a sequence from the min to the max year. This returns a named list which is fed to stack to produce the following.
values ind
1 2001 A
2 2002 A
3 2003 A
4 2004 A
5 2002 B
6 2003 B
7 2004 B
8 2005 B
If you are curious how you might do this in data.table, it's also pretty straight forward:
library(data.table)
setDT(df)[, .(year=min(year):max(year)), by=x]
which returns
x year
1: A 2001
2: A 2002
3: A 2003
4: A 2004
5: B 2002
6: B 2003
7: B 2004
8: B 2005