Statements (51)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
statistical analysis
|
| gptkbp:advantage |
Interpretability
Computational efficiency |
| gptkbp:assumes |
Linearity between input and output
|
| gptkbp:basisFor |
Many machine learning algorithms
|
| gptkbp:canBe |
Sparse
Multivariate Univariate Regularized Unsupervised Supervised |
| https://www.w3.org/2000/01/rdf-schema#label |
Linear models
|
| gptkbp:include |
gptkb:Perceptron
gptkb:Logistic_regression gptkb:Linear_regression gptkb:Ridge_regression gptkb:Lasso_regression gptkb:Elastic_net Support vector machine (linear kernel) |
| gptkbp:limitation |
Cannot model non-linear relationships
|
| gptkbp:mathematicallyExpressedAs |
y = Xw + b
|
| gptkbp:originatedIn |
19th century
|
| gptkbp:output |
Linear combination of inputs
|
| gptkbp:parameter |
Weights
Bias |
| gptkbp:relatedTo |
gptkb:Generalized_linear_models
Nonlinear models Polynomial regression |
| gptkbp:requires |
Homoscedasticity
Independence of errors No multicollinearity Feature scaling (sometimes) Normality of errors (for inference) |
| gptkbp:trainer |
Gradient descent
Least squares |
| gptkbp:usedFor |
Prediction
Regression Classification Feature selection |
| gptkbp:usedIn |
gptkb:signal_processing
Finance Time series analysis Biology Statistics Social sciences Data science Machine learning Econometrics |
| gptkbp:bfsParent |
gptkb:Linear_Models_in_Statistics
gptkb:Best_Linear_Unbiased_Estimator |
| gptkbp:bfsLayer |
8
|