fit_blended function

Fit a Blended mixture using an ECME-Algorithm

fit_blended(
  dist,
  obs,
  start,
  min_iter = 0L,
  max_iter = 100L,
  skip_first_e = FALSE,
  tolerance = 1e-05,
  trace = FALSE,
  ...
)

Arguments

• dist: A BlendedDistribution. It is assumed, that breaks and bandwidths are not a placeholder and that weights are to be estimated.
• obs: Set of observations as produced by trunc_obs() or convertible via as_trunc_obs().
• start: Initial values of all placeholder parameters. If missing, starting values are obtained from fit_dist_start().
• min_iter: Minimum number of EM-Iterations
• max_iter: Maximum number of EM-Iterations (weight updates)
• skip_first_e: Skip the first E-Step (update Probability weights)? This can help if the initial values cause a mixture component to vanish in the first E-Step before the starting values can be improved.
• tolerance: Numerical tolerance.
• trace: Include tracing information in output? If TRUE, additional tracing information will be added to the result list.
• ...: Passed to fit_dist_start() if start is missing.

Returns

A list with elements

• params the fitted parameters in the same structure as init.
• params_hist (if trace is TRUE) the history of parameters (after each e- and m- step)
• iter the number of outer EM-iterations
• logLik the final log-likelihood

Examples

dist <- dist_blended(
   list(
     dist_exponential(),
     dist_genpareto()
   )
 )

params <- list(
  probs = list(0.9, 0.1),
  dists = list(
    list(rate = 2.0),
    list(u = 1.5, xi = 0.2, sigmau = 1.0)
  ),
  breaks = list(1.5),
  bandwidths = list(0.3)
)

x <- dist$sample(100L, with_params = params)

dist$default_params$breaks <- params$breaks
dist$default_params$bandwidths <- params$bandwidths
if (interactive()) {
  fit_blended(dist, x)
}

See Also

Other distribution fitting functions: fit_dist(), fit_erlang_mixture(), fit_mixture()