Class NDPolyFitResult
- Namespace
- NumSharp
- Assembly
- NumSharp.dll
The result of a numpy.polynomial least-squares fit (polyfit, chebfit, legfit,
lagfit, hermfit, hermefit): the fitted coefficient array ITSELF — an NDArray, so
it indexes, computes and passes to every function (chebval(x, chebfit(x, y, 3))) exactly like NumPy's
ndarray — that also carries NumPy's full=True statistics.
public sealed class NDPolyFitResult : NDArray, IIndex, ICloneable, IEnumerable, IDisposable
- Inheritance
-
NDPolyFitResult
- Implements
- Inherited Members
- Extension Methods
Remarks
NumPy returns the coefficient ndarray, or with full=True the tuple (c, [resids, rank, sv,
rcond]). C# cannot change a method's return type with a flag, so the one result stands for both:
NDArray c = P.polyfit(x, y, 3) and var c = P.polyfit(x, y, 3) are the coefficients, and
var (c, (resids, rank, sv, rcond)) = P.polyfit(x, y, 3, full: true) deconstructs as NumPy's tuple (or
var (c, info) = …). Without full the four statistics are null and deconstruction is refused, as
NumPy's ndarray does not unpack into (c, info) either.
Lifetime. Disposing the result releases the coefficients only. The statistics are separate arrays the result owns: no NDScope tracks them (a scope that yields the result would yield only the coefficient array, and would reclaim scope-tracked statistics under it), so they are released by the garbage collector, or by disposing them explicitly.
Its memory is NumPy's: for a scalar degree the coefficients are a VIEW of the computed quotient (OWNDATA false), for a degree list a new array that owns its data.
Constructors
NDPolyFitResult(UnmanagedStorage, NDArray, NDArray, NDArray, NDArray)
Wraps the coefficient storage and the statistics.
public NDPolyFitResult(UnmanagedStorage storage, NDArray residuals, NDArray rank, NDArray singularValues, NDArray rcond)
Parameters
storageUnmanagedStorageThe coefficients' storage (an alias for a view, or the owner's storage taken over).
residualsNDArraylstsq's residuals, or null without
full.rankNDArraylstsq's rank (a 0-d int32, NumPy's np.intc), or null without
full.singularValuesNDArraylstsq's singular values, or null without
full.rcondNDArrayThe rcond used, or null without
full.
Remarks
The fit functions build this; a caller building one itself gets the same object. The four statistics are kept
by reference (not copied) and become the result's to own: disposing one of them while the result is still read
leaves residuals and its siblings pointing at a released buffer. Deconstruction reads
residuals as the full=True mark: null refuses it, so pass all four or none.
Exceptions
- NullReferenceException
storageisnull.
Properties
rank
The numerical rank of the scaled design matrix (a 0-d int32); null unless full: true.
public NDArray rank { get; }
Property Value
rcond
The rcond the fit used: the caller's value (an NDArray as given, a number as the 0-d array of its dtype), or the
default len(x)*eps of x's float dtype. Null unless full: true.
public NDArray rcond { get; }
Property Value
residuals
The sum of squared residuals of the fit, per data set (NumPy's resids): EMPTY unless the scaled design
matrix has full rank and more points than coefficients. Null unless the fit ran with full: true.
public NDArray residuals { get; }
Property Value
singular_values
The singular values of the scaled design matrix; null unless full: true.
public NDArray singular_values { get; }
Property Value
Methods
Deconstruct(out NDArray, out (NDArray residuals, NDArray rank, NDArray singular_values, NDArray rcond))
NumPy's c, [resids, rank, sv, rcond] = …fit(…, full=True): the coefficients and the four statistics.
public void Deconstruct(out NDArray c, out (NDArray residuals, NDArray rank, NDArray singular_values, NDArray rcond) info)
Parameters
cNDArrayThe coefficients (this array).
info(NDArray, NDArray, NDArray, NDArray)(residuals, rank, singular values, rcond) — NumPy's list, in its order.
Exceptions
- ValueError
The fit ran without
full(NumPy's result is the bare coefficient array, which does not unpack into(c, info)).