The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
Initial public release.
metahunt() chains denoised functional SPA basis
hunting, constrained simplex projection, and Dirichlet weight modelling
in a single call. Method dispatch for predict(),
summary(), and plot() on the returned
"metahunt" object.split_conformal() and cross_conformal()
return distribution-free prediction intervals around the target function
(pointwise on the grid or, with a wrapper, around a scalar
summary).conformal_from_fit() adds intervals to an already-fit
pipeline using a held-out calibration set.coverage(), summary(), and
plot() methods for the "metahunt_conformal"
class.reconstruction_error_curve() (unsupervised elbow) and
cv_error_curve() (supervised CV) for picking
K.select_denoising_params() cross-validates the
(N, Delta) knobs of dfspa().dfspa() denoised functional Successive Projection
Algorithm (Algorithm 1 of the paper).project_to_simplex() constrained simplex projection of
each study’s function onto the recovered bases (quadratic program via
quadprog).fit_weight_model() and
predict.metahunt_weight_model() for Dirichlet regression of
simplex weights on study-level covariates, with
coef.metahunt_weight_model() for inspecting
coefficients.predict_target() and apply_wrapper() for
composing predictions and scalar summaries by hand.build_grid() constructs a shared evaluation grid from
any reference patient-level dataset.f_hat_from_models() evaluates a list of fitted models
on the shared grid with class-aware dispatch for ranger,
grf (causal_forest,
regression_forest), and a default branch that covers
lm/glm/randomForest. Custom S4
classes can supply their own predict_fn.minmax_regret() implements the covariate-free
worst-case-regret aggregator of Zhang, Huang, and Imai (2024, arXiv:2412.11136).metahunt-intro, data-prep,
grid-weights, wrapper-scalar, plus
get-started.These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.