Multi-Metric Weighted kNN

Usage

predict_wknn_multi(
  train_data,
  test_data,
  train_labels,
  k = 5,
  method = "pearson",
  return_matrix = FALSE,
  chunk_size = 10000L
)

train_data: Data frame or matrix of training cells (rows = markers, cols = cells).
test_data: Data frame or matrix of test cells (rows = markers, cols = cells).
train_labels: Factor of labels for the training data.
k: Number of neighbors.
method: One of "pearson", "spearman", "cosine", or "euclidean"
return_matrix: Logical indicating whether to return the full probability matrix for all classes. If FALSE, only the predicted labels and associated probabilities are returned. Default is FALSE.
chunk_size: Optional integer number of test cells per block. If NULL, all test cells are processed in a single block (legacy behavior).

A list containing predicted 'labels' and 'probs'.