Development build. This page describes main, not a released package. The latest published Lodestar.Preprocessing is 0.1.0 — read its documentation.

Home › Preprocessing › Feature scaling

Maps a row-major sample matrix onto the fitted range.

public double[] Transform(ReadOnlySpan<double> samples)

Parameters — samples is the matrix to map, row-major, with FeatureCount values per row.

Returns — a new array of the same length.

Exceptions — ArgumentException when samples holds no row, a partial one, or a non-finite value.

Example — a value the fit never saw lands outside the range, unless clipping is asked for.

using Lodestar.Preprocessing;

double[] seen = [0.0, 10.0];

MinMaxScaler plain = MinMaxScaler.Fit(seen, featureCount: 1);
MinMaxScaler clipped = MinMaxScaler.Fit(seen, 1, new MinMaxScalerOptions { Clip = true });

double outside = plain.Transform([20.0])[0];    // => 2
double bounded = clipped.Transform([20.0])[0];  // => 1

Remarks — never writes to the input; scikit-learn's copy=False has no counterpart, since a span the caller owns is not this package's to overwrite.

Clipping belongs to this direction only. InverseTransform does not clip, which is the reference's asymmetry and not an oversight — see that page.

Applies to — net10.0, netstandard2.0.

See also — MinMaxScalerOptions, MinMaxScaler.InverseTransform.