nimare.meta.ibma
.VarianceBasedLikelihood
- class VarianceBasedLikelihood(method='ml', *args, **kwargs)[source]
Bases:
nimare.base.MetaEstimator
A likelihood-based meta-analysis method for estimates with known variances.
Changed in version 0.0.8:
[FIX] Remove single-dimensional entries of each array of returns (
dict
).
New in version 0.0.4.
Iteratively estimates the between-subject variance tau^2 and fixed effect coefficients using the specified likelihood-based estimator (ML or REML).
- Parameters
method ({‘ml’, ‘reml’}, optional) – The estimation method to use. Either ‘ml’ (for maximum-likelihood) or ‘reml’ (restricted maximum-likelihood). Default is ‘ml’.
Notes
Requires
beta
andvarcope
images.The ML and REML solutions are obtained via SciPy’s scalar function minimizer (
scipy.optimize.minimize()
). Parameters tominimize()
can be passed in as keyword arguments.Warning
Likelihood-based estimators are not parallelized across voxels, so this method should not be used on full brains, unless you can submit your code to a job scheduler.
Masking approaches which average across voxels (e.g., NiftiLabelsMaskers) will likely result in biased results. The extent of this bias is currently unknown.
All image-based meta-analysis estimators adopt an aggressive masking strategy, in which any voxels with a value of zero in any of the input maps will be removed from the analysis.
References
DerSimonian, R., & Laird, N. (1986). Meta-analysis in clinical trials. Controlled clinical trials, 7(3), 177-188.
Kosmidis, I., Guolo, A., & Varin, C. (2017). Improving the accuracy of likelihood-based inference in meta-analysis and meta-regression. Biometrika, 104(2), 489–496. https://doi.org/10.1093/biomet/asx001
See also
pymare.estimators.VarianceBasedLikelihoodEstimator
The PyMARE estimator called by this class.
- fit(dataset, drop_invalid=True)[source]
Fit Estimator to Dataset.
- Parameters
dataset (
nimare.dataset.Dataset
) – Dataset object to analyze.drop_invalid (
bool
, optional) – Whether to automatically ignore any studies without the required data or not. Default is False.
- Returns
nimare.results.MetaResult
– Results of Estimator fitting.- Variables
inputs_ (
dict
) – Inputs used in _fit.
Notes
The
fit
method is a light wrapper that runs input validation and preprocessing before fitting the actual model. Estimators’ individual “fitting” methods are implemented as_fit
, although users should callfit
.