Package index • MetaHunt

End-to-end fitter

One-call interface; covers the full pipeline.

metahunt(): Fit the full MetaHunt pipeline
predict(<metahunt>): Predict target functions (or scalar summaries) from a MetaHunt fit
summary(<metahunt>): Summarise a MetaHunt fit
print(<summary.metahunt>): Print a summary.metahunt object
plot(<metahunt>): Plot recovered basis functions from a MetaHunt fit

Distribution-free prediction intervals around the target function.

split_conformal(): Split conformal prediction intervals for target-function predictions
cross_conformal(): Cross-conformal prediction intervals (pooled K-fold scores)
conformal_from_fit(): Split conformal intervals from a pre-fit MetaHunt pipeline
coverage(): Empirical coverage of a conformal prediction-interval object
summary(<metahunt_conformal>): Summarise a conformal prediction-interval object
plot(<metahunt_conformal>): Plot a conformal prediction-interval object

Diagnostics for choosing the number of bases K and the d-fSPA denoising knobs.

reconstruction_error_curve(): Reconstruction-error curve for basis-rank selection
cv_error_curve(): Cross-validated prediction-error curve for basis-rank selection
select_denoising_params(): Choose d-fSPA denoising parameters by cross-validation
print(<metahunt_denoising_search>): Print method for d-fSPA denoising parameter search results

Lower-level primitives that can be composed independently.

dfspa(): Denoised functional Successive Projection Algorithm (d-fSPA)
project_to_simplex(): Project study-level functions onto the simplex spanned by basis functions
fit_weight_model(): Fit a weight model mapping study-level covariates to simplex weights
predict(<metahunt_weight_model>): Predict simplex weights for new study-level covariates
coef(<metahunt_weight_model>): Extract coefficients from a MetaHunt weight model
predict_target(): Predict the target function for new study-level covariates
apply_wrapper(): Reduce predicted functions to scalars via a user-supplied wrapper

Onramp from fitted-model lists to MetaHunt’s matrix inputs.

build_grid(): Build a shared evaluation grid from a reference dataset
f_hat_from_models(): Build the F_hat matrix from a list of fitted study-level models

Covariate-free worst-case-regret aggregator (Zhang, Huang & Imai 2024).

minmax_regret(): Minimax-regret aggregator for multisite function-valued estimands

MetaHunt MetaHunt-package: MetaHunt: Privacy-Preserving Meta-Analysis via Low-Rank Basis Hunting