The non-missing cases indicate the training set, and missing cases indicate the test set.

`impute_knn(x, data, seed = 123456)`

- x
clustering object

- data
data matrix

- seed
random seed for knn imputation reproducibility

An object with (potentially not all) missing values imputed with K-Nearest Neighbours.

We consider 5 nearest neighbours and the minimum vote for definite decision is 3.

Other imputation functions:
`impute_missing()`

```
data(hgsc)
dat <- hgsc[1:100, 1:50]
x <- consensus_cluster(dat, nk = 4, reps = 4, algorithms = c("km", "hc",
"diana"), progress = FALSE)
x <- apply(x, 2:4, impute_knn, data = dat, seed = 1)
```