How to use approxExtrap (Hmisc package) with dplyr?

Question

I'm trying to perform linear extrapolation and can't figure out the correct notation to apply the approxExtrap function from the Hmisc package. I have seen some examples, but none that I've been able to apply. I have been able to use the normal approx function specifying only the "x" argument, being the variable I want to interpolate. Doing the following leaves me with an error. Any suggestions? Thank you!

library(tidyverse)
library(zoo)
library(Hmisc)

#write data frame
df <- tibble(day=1:10,
             sales =  c(NA, NA, NA, 4, 5, 6, 7, 8 , 9, 10))

#attempt to perform extrapolation
df <- df %>% 
  mutate(sales=approxExtrap(x=sales, y=NULL)) 

Error in `mutate()`:
! Problem while computing `sales = approxExtrap(x = sales, y = NULL)`.
Caused by error in `approx()`:
! need at least two non-NA values to interpolate

score 0 · Answer 1 · answered Jul 31 '22 at 16:40

Your x should be the day column and y should be the column you want to extrapolate on. By saying xout=c(1:10) you specify the area you want to extrapolate on. Also the output of approxExtrap is a list of the output from x and y, so you have to keep that in mind. You can use the following code:

library(tidyverse)
library(zoo)
library(Hmisc)

#write data frame
df <- tibble(day=1:10,
             sales =  c(NA, NA, NA, 4, 5, 6, 7, 8 , 9, 10))

#attempt to perform extrapolation
df %>% 
  mutate(sales=approxExtrap(x=day, y=sales, xout = c(1:10), method = "linear")$y) 
#> # A tibble: 10 × 2
#>      day sales
#>    <int> <dbl>
#>  1     1     4
#>  2     2     4
#>  3     3     4
#>  4     4     4
#>  5     5     5
#>  6     6     6
#>  7     7     7
#>  8     8     8
#>  9     9     9
#> 10    10    10

^{Created on 2022-07-31 by the reprex package (v2.0.1)}

Thank you for your answer. What does adding $y do? The output does not seem accurate, since days 1, 2, and 3 are assigned a sales value of 4. — dd_data, Aug 01 '22 at 16:33

score 0 · Answer 2 · answered Mar 21 '23 at 08:48

The reply by Quinten was almost there. It was missing the identification where the data has NA values. This should work:

 #write data frame
df <- tibble(day=1:10,
             sales =  c(NA, NA, NA, 40, 50, 60, 70, 80 , 90, 100))

#attempt to perform extrapolation
df %>% 
  mutate(sales=approxExtrap(x=day[!is.na(sales)], y=sales[!is.na(sales)], xout = c(1:10), method = "linear")$y)

# A tibble: 10 × 2
     day sales
   <int> <dbl>
 1     1    10
 2     2    20
 3     3    30
 4     4    40
 5     5    50
 6     6    60
 7     7    70
 8     8    80
 9     9    90
10    10   100

Adding $y returns the wanted part of the list. See also: R linear extrapolate missing values

How to use approxExtrap (Hmisc package) with dplyr?

2 Answers2