details_discrim_linear_sda {parsnip} | R Documentation |
Linear discriminant analysis via James-Stein-type shrinkage estimation
Description
sda::sda()
can fit a linear discriminant analysis model that can fit models
between classical discriminant analysis and diagonal discriminant analysis.
Details
For this engine, there is a single mode: classification
Tuning Parameters
This engine has no tuning parameter arguments in
discrim_linear()
.
However, there are a few engine-specific parameters that can be set or
optimized when calling set_engine()
:
-
lambda
: the shrinkage parameters for the correlation matrix. This maps to the parameterdials::shrinkage_correlation()
. -
lambda.var
: the shrinkage parameters for the predictor variances. This maps todials::shrinkage_variance()
. -
lambda.freqs
: the shrinkage parameters for the class frequencies. This maps todials::shrinkage_frequencies()
. -
diagonal
: a logical to make the model covariance diagonal or not. This maps todials::diagonal_covariance()
.
Translation from parsnip to the original package
The discrim extension package is required to fit this model.
library(discrim) discrim_linear() %>% set_engine("sda") %>% translate()
## Linear Discriminant Model Specification (classification) ## ## Computational engine: sda ## ## Model fit template: ## sda::sda(Xtrain = missing_arg(), L = missing_arg(), verbose = FALSE)
Preprocessing requirements
Factor/categorical predictors need to be converted to numeric values
(e.g., dummy or indicator variables) for this engine. When using the
formula method via fit()
, parsnip will
convert factor columns to indicators.
Variance calculations are used in these computations so zero-variance predictors (i.e., with a single unique value) should be eliminated before fitting the model.
Case weights
The underlying model implementation does not allow for case weights.
References
Ahdesmaki, A., and K. Strimmer. 2010. Feature selection in omics prediction problems using cat scores and false non-discovery rate control. Ann. Appl. Stat. 4: 503-519. Preprint.