bootstrap_model()

bootstrap_model() resamples sequences with replacement, refits the model on each bootstrap sample, and collects parameter estimates for uncertainty quantification.

Function Usage

bootstrap_model(
    model,
    n_sim=100,
    method="nonparametric",
    random_state=None,
    verbose=True,
    n_jobs=1
)

seqHMM Parameter Mapping

Sequenzo	seqHMM bootstrap_coefs()
model	Fitted hmm, mhmm, or nhmm
n_sim	Number of bootstrap replicates
method="nonparametric"	Resample sequences (nonparametric bootstrap)
Return summary	Percentile confidence intervals

Entry Parameters

Parameter	Required	Type	Description
model	✓	HMM / MHMM / NHMM	Fitted model.
n_sim	✗	int	Bootstrap replicates. Default 100.
method	✗	str	"nonparametric" (default). "parametric" not yet implemented.
random_state	✗	int / None	RNG seed.
verbose	✗	bool	Show progress bar when tqdm is available. Default True.
n_jobs	✗	int	Parallel jobs (not yet implemented; always 1).

What It Returns

A dict:

Key	Description
bootstrap_samples	List of parameter dicts per replicate
original_model	Input fitted model
n_sim, method	Settings used
summary	Means, standard deviations, and percentile CIs (e.g. ci_95)

Example

from sequenzo.seqhmm import build_hmm, fit_model, bootstrap_model

hmm = build_hmm(seq, n_states=4, random_state=42)
hmm = fit_model(hmm)

boot = bootstrap_model(hmm, n_sim=100, verbose=True)
ci = boot["summary"]["initial_probs"]["ci_95"]
print(ci)

R Counterpart

• Closest R function: seqHMM bootstrap_coefs()
• Mapping note: R's function targets NHMM/MNHMM coefficients; Sequenzo bootstraps full refits for HMM, MHMM, and NHMM.

Notes

• Computationally expensive: each replicate refits the full model.
• Failed refits on a replicate are skipped silently in the bootstrap loop.
• Use moderate n_sim (50–200) for exploratory intervals; increase for publication-grade precision.

Authors

Code: Yuqi Liang, Yapeng Wei

Documentation: Yuqi Liang

References

Helske, S., & Helske, J. (2019). Mixture hidden Markov models for sequence data: The seqHMM package in R. Journal of Statistical Software, 88(3), 1–32.