Parametric family fitting, transform metadata, and the unified distribution selector used by the pipeline.

Package source: tabnetics.distribution

Package overview

Distribution fitting and transform helpers.

Stable exports

  • class BootstrapGOFSelector (class) - Source. Monte-Carlo goodness-of-fit selector tailored to small samples.
  • def select_best_distribution_bootstrap(data: np.ndarray, distributions: Dict[str, sps.rv_continuous] | None = None, statistic: str = 'cvm', n_boot: int = 2000, random_state: Any = None) -> Tuple[str, List[Dict[str, Any]]] (function) - Source. Return (best_dist_name, sorted_result_list).
  • def select_best_distribution_bic(data: np.ndarray, distributions: Dict[str, sps.rv_continuous] | None = None) -> Tuple[str, List[Dict[str, Any]]] (function) - Source. Return (best_dist_name, sorted_result_list) using BIC.
  • def auto_select_distribution(data: np.ndarray, *, small_sample_threshold: int = 30, statistic: str = 'cvm', n_boot: int = 2000, random_state: Any = None) -> Tuple[str, List[Dict[str, Any]]] (function) - Source. Dispatch to the most appropriate selection strategy.
  • class DistributionFeatures (class) - Source. Represents statistical features extracted from data, used to guide distribution selection. These features help in applying heuristics and bonuses for more accurate fitting.
  • class TransformInfo (class) - Source. Stores information about data transformations (shifting, scaling) applied prior to fitting a distribution. Used to reverse-transform fitted parameters.
  • class LRTResult (class) - Source. Stores results of Likelihood Ratio Test between nested distributions.
  • class CVResult (class) - Source. Stores cross-validation results for a distribution.
  • class FitResult (class) - Source. Holds the results of fitting a single distribution to data, including parameters, goodness-of-fit statistics, and any applied transformations.
  • class UnifiedDistributionSelectorV6 (class) - Source. Hybrid distribution selector combining statistical fitting, goodness-of-fit tests, and feature-based heuristics to identify the best-fitting distribution for given data. This version incorporates improvements for robustness and accuracy, including LRT and CV.

Module details

tabnetics.distribution.__init__

Source file

Distribution fitting and transform helpers.

No top-level public symbols are exported directly from this module.

tabnetics.distribution.bootstrap

Source file

  • class BootstrapGOFSelector (class) - Source. Monte-Carlo goodness-of-fit selector tailored to small samples.
  • def select_best_distribution_bootstrap(data: np.ndarray, distributions: Dict[str, sps.rv_continuous] | None = None, statistic: str = 'cvm', n_boot: int = 2000, random_state: Any = None) -> Tuple[str, List[Dict[str, Any]]] (function) - Source. Return (best_dist_name, sorted_result_list).
  • def select_best_distribution_bic(data: np.ndarray, distributions: Dict[str, sps.rv_continuous] | None = None) -> Tuple[str, List[Dict[str, Any]]] (function) - Source. Return (best_dist_name, sorted_result_list) using BIC.
  • def auto_select_distribution(data: np.ndarray, *, small_sample_threshold: int = 30, statistic: str = 'cvm', n_boot: int = 2000, random_state: Any = None) -> Tuple[str, List[Dict[str, Any]]] (function) - Source. Dispatch to the most appropriate selection strategy.

tabnetics.distribution.selector

Source file

  • class DistributionFeatures (class) - Source. Represents statistical features extracted from data, used to guide distribution selection. These features help in applying heuristics and bonuses for more accurate fitting.
  • class TransformInfo (class) - Source. Stores information about data transformations (shifting, scaling) applied prior to fitting a distribution. Used to reverse-transform fitted parameters.
  • class LRTResult (class) - Source. Stores results of Likelihood Ratio Test between nested distributions.
  • class CVResult (class) - Source. Stores cross-validation results for a distribution.
  • class FitResult (class) - Source. Holds the results of fitting a single distribution to data, including parameters, goodness-of-fit statistics, and any applied transformations.
  • class UnifiedDistributionSelectorV6 (class) - Source. Hybrid distribution selector combining statistical fitting, goodness-of-fit tests, and feature-based heuristics to identify the best-fitting distribution for given data. This version incorporates improvements for robustness and accuracy, including LRT and CV.

Documentation and webpages on this site are generated from authoritative internal sources using a combination of deterministic rules and generative AI. Errors are possible. Please report issues via GitHub Discussions or email [email protected].