Plot data embedding with SmoothEM component means — plot_EM

Visualize an EM fit (from EM_algorithm()) on 1D or 2D selected coordinates. For 2D, draws arrows connecting component means in order. For 1D, draws a scatter of posterior position vs the selected coordinate and overlays the mean curve.

plot_EM_embedding(
  fit,
  X,
  dims = c(1, 2),
  position = NULL,
  use_posterior_mean = TRUE,
  pch = 19,
  col = "grey60",
  cex = 0.7,
  mu_pch = 8,
  mu_col = "orange",
  mu_cex = 1,
  arrow_col = "orange",
  arrow_lwd = 3,
  arrow_len = 0.08,
  line_col = "orange",
  line_lwd = 2,
  add = FALSE,
  xlab = NULL,
  ylab = NULL,
  main = NULL,
  ...
)

Arguments

fit: A fitted object returned by EM_algorithm().
X: Numeric matrix (n x d). The data used to fit the model.
dims: Integer vector of length 1 or 2 indicating which coordinates (columns of X) to plot.
position: Optional numeric vector of length n giving x-axis "position". If NULL, uses fit$position if present; otherwise uses posterior mean position computed as fit$gamma %*% seq_len(K).
use_posterior_mean: Logical; if TRUE (default), use posterior mean position. If FALSE, use MAP position via max.col(fit$gamma).
pch, col, cex: Point style for data scatter.
mu_pch, mu_col, mu_cex: Mean point style.
arrow_col, arrow_lwd, arrow_len: Arrow style (2D only).
line_col, line_lwd: Mean curve style (1D only).
add: Logical; if TRUE, add to existing plot.
xlab, ylab, main: Labels; if NULL, auto-generated.
...: Passed to plot() for the scatter.

Value

Invisibly returns a list with dims, pos, mu_mat.