在 R 中按组用 NA 填充缺失的日期 - 在日期范围的末尾也有 NA

Question

I would like to generate empty rows (NAs) for missing dates in a large dataset. For context, this is a large dataset where each individual (ID) has various years of data.

Here is a simplified version of the data for two individuals:

table <- "ID    Date    dist.km
 1 1     2007-10-15     15147
 2 1     2007-10-16     15156
 3 1     2007-10-17     15173
 4 1     2007-10-18     15185
 5 1     2007-10-19     15194
 6 1     2007-10-25     15202
 7 1     2007-10-26     15216
 8 1     2007-10-27     15240
 9 1     2007-10-28     15270
10 1     2007-10-29     15290
11 2     2008-10-15     15147
12 2     2008-10-16     15156
13 2     2008-10-17     15173
14 2     2008-10-18     15185
15 2     2008-10-19     15194
16 2     2008-10-20     15202
17 2     2008-10-21     15216
18 2     2008-10-29     15240
19 2     2008-10-30     15270
20 2     2008-10-31     15290"

#Create a dataframe with the above table
df <- read.table(text=table, header = TRUE)
df

I first tried using complete() , here:

library(tidyverse)

newdat <- complete(df, ID, Date)
newdat

The output of this is the original dataset, with NA values for all dates outside of the dataset date range. So, some dates are not filled in. For example, 2007-10-20 to 2007-10-24 did not fill in for ID 1. So essentially it is filling in NA values for dates outside of my date range of distance data, but not within it.

I then tried this format:

library(dplyr)
library(tidyr)

newdat2 <- dat %>% group_by(ID) %>%
  complete(Date = seq.Date(min(Date), max(Date), by = "day"))
newdat2

And this worked, but this does not produce dates outside of the date range of each ID. So each format produces the opposite results. I am looking to get at least one NA at the end of each ID's date range to show the end of this set. If this can't be done using complete() , maybe my question is: how can I add a blank NA cell for a specific date in every year of my dataset? All datasets start on 10-15 and end on 02-15. So, how can I add one NA cell for 02-16 for each ID in each year?

Any help would be appreciated.

Answer 1

I think you're close with your second attempt. If you want to manually enforce the limits of the expansion in the complete call, you can do it there. It wasn't clear what limits you were after but perhaps the below can get you there. Note that I used two date ranges because it seemed like you wanted to hit two time ranges. But adjust if I misunderstood. Can also be called programmatically if you have those dates stored somewhere. Also, I converted your date column to an actual date format using as.Date() during import.

library(tidyverse)

table <- "ID    Date    dist.km\n 1 1     2007-10-15     15147\n 2 1     2007-10-16     15156\n 3 1     2007-10-17     15173\n 4 1     2007-10-18     15185\n 5 1     2007-10-19     15194\n 6 1     2007-10-25     15202\n 7 1     2007-10-26     15216\n 8 1     2007-10-27     15240\n 9 1     2007-10-28     15270\n10 1     2007-10-29     15290\n11 2     2008-10-15     15147\n12 2     2008-10-16     15156\n13 2     2008-10-17     15173\n14 2     2008-10-18     15185\n15 2     2008-10-19     15194\n16 2     2008-10-20     15202\n17 2     2008-10-21     15216\n18 2     2008-10-29     15240\n19 2     2008-10-30     15270\n20 2     2008-10-31     15290"

#Create a dataframe with the above table
df <- read.table(text=table, header = TRUE) %>% 
  mutate(Date = as.Date(Date))

# expand by feeding the limits of the date ranges to cover
newdat2 <- df %>%
  group_by(ID) %>%
  complete(Date = c(
    seq.Date(
      from = as.Date("2007-10-15"),
      to = as.Date("2008-02-15"),
      by = "day"
    ),
    seq.Date(
      from = as.Date("2008-10-15"),
      to = as.Date("2009-02-15"),
      by = "day"
    )
  ))

newdat2

#> # A tibble: 496 x 3
#> # Groups:   ID [2]
#>       ID Date       dist.km
#>    <int> <date>       <int>
#>  1     1 2007-10-15   15147
#>  2     1 2007-10-16   15156
#>  3     1 2007-10-17   15173
#>  4     1 2007-10-18   15185
#>  5     1 2007-10-19   15194
#>  6     1 2007-10-20      NA
#>  7     1 2007-10-21      NA
#>  8     1 2007-10-22      NA
#>  9     1 2007-10-23      NA
#> 10     1 2007-10-24      NA
#> # ... with 486 more rows

^{Created on 2021-03-15 by the reprex package (v1.0.0)}