lib5c.algorithms.variance.mle module¶
-
lib5c.algorithms.variance.mle.
mle_variance
(obs, exp, model='lognorm', min_obs=2, min_dist=6, regional=False)[source]¶ Fits a single point estimate of the dispersion across each or all regions under the selected model, and returns the converted variance estimates.
Parameters: - exp (obs,) – The counts dicts of observed and expected data. Keys are region names, values are square, symmetric count matrices.
- model ({'lognorm', 'loglogistic', 'nbinom'}) – The statistical model to use for MLE point estimation.
- min_obs (float) – Fit only points with at least this many observed counts.
- min_dist (int) – Fit only points with at least this interaction distance in bin units.
- regional (bool) – Pass True to fit a separate point estimate for each region.
Returns: The variance estimates.
Return type: dict of np.ndarray