Preprocess a weight matrix for downstream steps

Preprocess a weight matrix for downstream steps

Usage

prepare_weight_matrix(
  W,
  adjust = TRUE,
  top_v = NULL,
  row_normalize = TRUE,
  zero_diag = TRUE,
  eps = 1e-08,
  keep_ties = TRUE
)

adjust_weight_matrix(W, zero_diag = TRUE, eps = 1e-08)

truncate_top_v_rows(W, top_v = NULL, keep_ties = TRUE, sparse = FALSE)

row_normalize_weights(W, eps = 1e-12)

Arguments

W

A square numeric weight matrix.

adjust

Logical; whether to apply adjusted-weight scaling.

top_v

Optional integer. If set, each row keeps only the top-v entries.

row_normalize

Logical; whether to row-normalize after adjustment/truncation.

zero_diag

Logical; whether to set diagonal to zero.

eps

Small positive value used to avoid division by zero.

keep_ties

Logical; whether top-v truncation keeps ties at the cutoff.

sparse

Logical; whether truncate_top_v_rows() should return a sparse matrix.

Value

A processed square numeric matrix.

Adjusted weight matrix using row-wise scaling by 1 - diag(W).

Top-v truncated matrix (row-wise).

Row-normalized weight matrix.