Inference and Estimation of Hidden Markov Models and Hidden Semi-Markov Models
Plot Conditional Return Levels from GEV-HMM
Plot Conditional Return Levels from GEV-HSMM
Bootstrap Confidence Intervals for Hidden Markov Models
Bootstrap Confidence Intervals for Hidden Semi-Markov Models
Plot Exceedance Probabilities from GEV-HMM
Plot Exceedance Probabilities from GEV-HSMM
Maximum Likelihood Estimation for Hidden Markov Models
Multiple Initialization Maximum Likelihood Estimation for Hidden Marko...
Maximum Likelihood Estimation for Hidden Semi-Markov Models
Fit Hidden Semi-Markov Model (HSMM) Without User-Provided Starting Val...
Generate Data from a Hidden Markov Model
Generate Data from a Hidden Semi-Markov Model
Global Decoding for Hidden Markov Models
Global Decoding of Hidden Semi-Markov Models
Variance-Covariance Matrix for Hidden Markov Models
Variance-Covariance Matrix for Hidden Semi-Markov Models
Calculate Information Criteria for HMM/HSMM Models
Local Decoding for Hidden Markov Models
Local Decoding for Hidden Semi-Markov Models
Plot Hidden Markov Model Parameters Over Time
Plot Hidden Semi-Markov Model Parameters Over Time
Ordinary Residuals for Hidden Markov Models
Ordinary Residuals for Hidden Semi-Markov Models
Provides flexible maximum likelihood estimation and inference for Hidden Markov Models (HMMs) and Hidden Semi-Markov Models (HSMMs), as well as the underlying systems in which they operate. The package supports a wide range of observation and dwell-time distributions, offering a flexible modelling framework suitable for diverse practical data. Efficient implementations of the forward-backward and Viterbi algorithms are provided via 'Rcpp' for enhanced computational performance. Additional functionality includes model simulation, residual analysis, non-initialised estimation, local and global decoding, calculation of diverse information criteria, computation of confidence intervals using parametric bootstrap methods, numerical covariance matrix estimation, and comprehensive visualisation functions for interpreting the data-generating processes inferred from the models. Methods follow standard approaches described by Guédon (2003) <doi:10.1198/1061860032030>, Zucchini and MacDonald (2009, ISBN:9781584885733), and O'Connell and Højsgaard (2011) <doi:10.18637/jss.v039.i04>.