kNN tricks & tips I: weighting donors
A variation of kNN imputation that is frequently applied uses the so-called distance-weighted aggregation. What this means is that when we aggregate the values from the neighbors to obtain a replacement for a missing value, we do so using the weighted mean and the weights are inverted distances from each neighbor. As a result, closer neighbors have more impact on the imputed value.
In this exercise, you will apply the distance-weighted aggregation while imputing the tao
data. This will only require passing two additional arguments to the kNN()
function. Let's try it out!
This exercise is part of the course
Handling Missing Data with Imputations in R
Exercise instructions
- Load the
VIM
package. - Impute
humidity
with kNN using distance-weighted mean for aggregating neighbors; you will need to specify thenumFun
andweightDist
arguments. - The margin plot to view the results has been already coded for you.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Load the VIM package
___(___)
# Impute humidity with kNN using distance-weighted mean
tao_imp <- ___(tao,
k = 5,
variable = "humidity",
___ = ___,
___ = ___)
tao_imp %>%
select(sea_surface_temp, humidity, humidity_imp) %>%
marginplot(delimiter = "imp")