|
gptkbp:instanceOf
|
gptkb:statistical_technique
gptkb:regression_analysis_method
|
|
gptkbp:adds
|
L2 penalty
|
|
gptkbp:alsoKnownAs
|
gptkb:Tikhonov_regularization
|
|
gptkbp:appliesTo
|
gptkb:generalized_linear_models
linear regression
|
|
gptkbp:assumes
|
linear relationship
|
|
gptkbp:biasVarianceTradeoff
|
increases bias, reduces variance
|
|
gptkbp:category
|
gptkb:machine_learning
statistics
linear models
|
|
gptkbp:citation
|
gptkb:Hoerl,_A._E.,_&_Kennard,_R._W._(1970)._Ridge_regression:_Biased_estimation_for_nonorthogonal_problems._Technometrics,_12(1),_55-67.
|
|
gptkbp:doesNotPerform
|
variable selection
|
|
gptkbp:effectOfLargeLambda
|
shrinks coefficients toward zero
|
|
gptkbp:introduced
|
gptkb:Andrey_Tikhonov
|
|
gptkbp:introducedIn
|
1970
|
|
gptkbp:lambda
|
regularization parameter
|
|
gptkbp:notRecommendedFor
|
highly sparse solutions
|
|
gptkbp:penaltyTerm
|
lambda times sum of squared coefficients
|
|
gptkbp:popularizedBy
|
gptkb:Arthur_E._Hoerl
gptkb:Robert_W._Kennard
|
|
gptkbp:reduces
|
sum of squared residuals plus penalty
variance of estimates
|
|
gptkbp:regularizes
|
least squares estimates
|
|
gptkbp:relatedTo
|
gptkb:ordinary_least_squares
gptkb:Lasso_regression
gptkb:Elastic_net
|
|
gptkbp:software
|
gptkb:SAS
gptkb:MATLAB
gptkb:scikit-learn
R
|
|
gptkbp:solutionFormula
|
(X^T X + λI)^{-1} X^T y
|
|
gptkbp:solvedBy
|
overfitting
closed-form
|
|
gptkbp:usedFor
|
gptkb:statistical_analysis
multicollinearity
|
|
gptkbp:bfsParent
|
gptkb:Lasso_regression
|
|
gptkbp:bfsLayer
|
7
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
Ridge regression
|