Mean squared error of a split at a single candidate partition point

Fits the growth-curve ensemble separately to the two sides of a candidate split d and returns the pooled MSE, as used by the grid search in find_partition_point().

Usage

split_mse(t, y, d, criterion = "mse", beta_fixed = 1.5)

Arguments

t, y

Time index and cumulative counts of the (open) window being searched, both starting at 1.

d

Candidate partition point, 1 <= d < length(y).

criterion, beta_fixed

Passed to fit_ensemble().

Value

A list with mse (pooled MSE, NA if either side fails to converge) and fit1, fit2 (the two ensemble_fit objects, or NULL on failure).