ldhmm-class: The ldhmm class

In ldhmm: Hidden Markov Model for Financial Time-Series Based on Lambda Distribution

Description

This S4 class is the major object class for ldhmm package

Slots

call

The match.call slot

m

numeric, length 1, number of states

param.nbr

numeric, number of parameters (2 or 3) for each ecld object

param

matrix, natural parameters for ecld objects, size of states times param.nbr. Each row can be 2-parameter sequences, or 3-parameter sequences. Three-parameter unit (mu, sigma, lambda) forms an ecld object representing a leptokurtic symmetric lambda distribution. On the other hand, to provide compatibility to a normal distribution HMM, two-parameter unit (mu, sigma) forms an ecld object with lambda=1.

gamma

matrix, the transition probability matrix, must be m by m.

delta

numeric, the initial distribution for each state, default is NULL.

stationary

logical, specify whether the initial distribution is stationary or not, default is TRUE.

mle.optimizer

character, the MLE optimizer. Currently it is just set to "nlm".

return.code

numeric, the return code from the MLE optimizer.

iterations

numeric, number of iterations MLE optimizer takes.

mllk

numeric, the final mllk value.

AIC

numeric, the final AIC.

BIC

numeric, the final BIC.

observations

numeric, stores the observations post optimization

states.prob

matrix, stores the state probabilities post optimization

states.local

numeric, stores the local decoding states post optimization

states.global

numeric, stores the global decoding states post optimization (Viterbi)

states.local.stats

matrix, stores the statistics of local states post optimization

states.global.stats

matrix, stores the statistics of global states post optimization

ldhmm documentation built on Jan. 11, 2020, 9:16 a.m.