CODEWITHSUNDEEP
rdplyrtibblelinear-interpolation
Aku-Ville Lehtimäki
Aku-Ville Lehtimäki

Reputation: 653

How to linearly interpolate a missing row in R?

Let's assume I have a following data:

library(dplyr)
mydata <- expand_grid(
    sex = c("male","female"),
    age = 20:30,
    employed = c(T,F),
    value1 = (1:5)^2,
    value2 = sqrt(1:5)) %>%
mutate(row_num = row_number()) %>%
select(row_num, everything())

Which would look like this:

# A tibble: 1,100 × 6
   row_num sex     age employed value1 value2
     <int> <chr> <int> <lgl>     <dbl>  <dbl>
 1       1 male     20 TRUE          1   1   
 2       2 male     20 TRUE          1   1.41
 3       3 male     20 TRUE          1   1.73
 4       4 male     20 TRUE          1   2   
 5       5 male     20 TRUE          1   2.24
 6       6 male     20 TRUE          4   1   
 7       7 male     20 TRUE          4   1.41
 8       8 male     20 TRUE          4   1.73
 9       9 male     20 TRUE          4   2   
10      10 male     20 TRUE          4   2.24

Let's now assume the row number three and eight went missing for some reason, i.e.

mydata %>% slice(-3,-8)
# A tibble: 1,098 × 6
   row_num sex     age employed value1 value2
     <int> <chr> <int> <lgl>     <dbl>  <dbl>
 1       1 male     20 TRUE          1   1   
 2       2 male     20 TRUE          1   1.41
 3       4 male     20 TRUE          1   2   
 4       5 male     20 TRUE          1   2.24
 5       6 male     20 TRUE          4   1   
 6       7 male     20 TRUE          4   1.41
 7       9 male     20 TRUE          4   2   
 8      10 male     20 TRUE          4   2.24
 9      11 male     20 TRUE          9   1   
10      12 male     20 TRUE          9   1.41

...but there could be missing rows in other categories (age, sex, employed) as well.

How could I linearly interpolate the missing rows (namely values for columns value1 and value2) and make mydata complete again? In other words I would like to use the information from rows 2 and 4 for row 3 AND information from rows 7 and 9 for row 8, and calculate the mean.

(I am pretty much aware that the logic behind columns value1 and value2 was not initially linear at all.)

Upvotes: 0

Views: 43

Answers (0)

Related Questions