SPLSHyperparameters

Properties

k

integer | object≥ 1

Number of latent components. Cannot exceed the number of features in the training data.

eta

number | object≥ 0< 1

Sparsity threshold: higher values select fewer features.

kappa

number | object≥ 0≤ 0.5

Concavity of the surrogate direction vector problem. Used with a multivariate coded outcome, i.e. multiclass classification.

select

string

Feature selection algorithm (regression only).

one of"pls2""simpls"

fit

string

PLS algorithm used for model fitting (regression only).

one of"kernelpls""widekernelpls""simpls""oscorespls"

classifier

string

Classifier fit on the latent components (classification only).

one of"lda""logistic"

scale_x

boolean

Scale features to unit variance.

scale_y

boolean

Scale the outcome to unit variance (regression only).

eps

number> 0

Convergence tolerance (regression only).

maxstep

integer≥ 1

Maximum number of iterations per component (regression only).

ifw

boolean | object

Inverse Frequency Weighting in classification.

Relationships