Crate pcw_regrs

Modules

• affine_min
  This module implements an efficient algorithm for determining the pointwise minimum of a collection of affine linear functions on ℝ.
• solve_dp
  The part of the solution concerned with the optimal partitions and general implementation of the dynamic program.

Macros

• dof
• max
  Find maximum of given values
• min
  Find minimum of given values

Structs

• DegreeOfFreedom
• MatchedUserParams
• ScoredModel
  A model for a timeseries and its CV score.
• SegmentModelSpec
  A specification of a model on some segment of data given by the start and stop indices w.r.t. a timeseries.
• Solution
  A solution of the optimization problem providing an interface to find the globally minimizing model of the CV score, the OSE optimal model and to investigate the CV and model functions.
• TimeSeriesSample
• UserParams
• ValidTimeSeriesSample
  A nonempty, NaN-free time series sample

Enums

• FitError
  The various kinds of errors that can happen during the polynomial fitting process.

Traits

• OrdFloat
  Safety
• RawFloat
  A marker trait that we use to wrap all the properties we internally need of a floating point type

Functions

• euclid_sq_metric
  Squared euclidean metric d(x,y)=‖x-y‖₂².
• try_fit_pcw_poly
  Fit a piecewise polynomial to the timeseries sample.

Type Aliases

• CvFunc
  A function mapping hyperparameter values to CV scores; each score being annotated with its standard error.
• CvMinimizerModel
  An optimal model minimizing the CV score.
• ModelFunc
  Maps each penalty γ to the corresponding optimal models given as piecewise model specifications: the arguments of the “inner” piecewise function are jump indices.
• OseBestModel
  An optimal model with respect to the one standard error rule.
• ScoredModelFunc
  Maps each penalty γ to the corresponding optimal model.