Generalized Linear Regression

Regression models for limited responses

For greater accuracy and link-function choices on low- through medium-dimensional data sets, fit a generalized linear model using fitglm.

For reduced computation time on high-dimensional data sets, train a binary, linear classification model, such as a logistic regression model, using fitclinear. You can also efficiently train a multiclass error-correcting output codes (ECOC) model composed of logistic regression models using fitcecoc.

For nonlinear classification with big data, train a binary, Gaussian kernel classification model with logistic regression using fitckernel.


GeneralizedLinearModelGeneralized linear regression model class
CompactGeneralizedLinearModelCompact generalized linear regression model class
ClassificationLinearLinear model for binary classification of high-dimensional data
ClassificationECOCMulticlass model for support vector machines (SVMs) and other classifiers
ClassificationKernelGaussian kernel classification model using random feature expansion
ClassificationPartitionedLinearCross-validated linear model for binary classification of high-dimensional data
ClassificationPartitionedLinearECOCCross-validated linear error-correcting output codes model for multiclass classification of high-dimensional data


fitglmCreate generalized linear regression model
stepwiseglmCreate generalized linear regression model by stepwise regression
compactCompact generalized linear regression model
dispDisplay generalized linear regression model
fevalEvaluate generalized linear regression model prediction
predictPredict response of generalized linear regression model
randomSimulate responses for generalized linear regression model
fitclinearFit linear classification model to high-dimensional data
templateLinearLinear classification learner template
fitcecocFit multiclass models for support vector machines or other classifiers
predictPredict labels for linear classification models
fitckernelFit Gaussian kernel classification model using random feature expansion
predictPredict labels for Gaussian kernel classification model
mnrfitMultinomial logistic regression
mnrvalMultinomial logistic regression values
glmfitGeneralized linear model regression
glmvalGeneralized linear model values
plotPartialDependenceCreate partial dependence plot (PDP) and individual conditional expectation (ICE) plots

Examples and How To

Generalized Linear Model Workflow

Fit a generalized linear model and analyze the results.

Train Logistic Regression Classifiers Using Classification Learner App

Create and compare logistic regression classifiers, and export trained models to make predictions for new data.

Fitting Data with Generalized Linear Models

This example shows how to fit and evaluate generalized linear models using glmfit and glmval.

Bayesian Analysis for a Logistic Regression Model

This example shows how to make Bayesian inferences for a logistic regression model using slicesample.


Generalized Linear Models

Generalized linear models use linear methods to describe a potentially nonlinear relationship between predictor terms and a response variable.

Multinomial Models for Nominal Responses

A nominal response variable has a restricted set of possible values with no natural order between them. A nominal response model explains and predicts the probability that an observation is in each category of a categorical response variable.

Multinomial Models for Ordinal Responses

An ordinal response variable has a restricted set of possible values that fall into a natural order. An ordinal response model describes the relationship between the cumulative probabilities of the categories and predictor variables.

Hierarchical Multinomial Models

A hierarchical multinomial response variable (also known as a sequential or nested multinomial response) has a restricted set of possible values that fall into hierarchical categories. The hierarchical multinomial regression models are extensions of binary regression models based on conditional binary observations.

Wilkinson Notation

Wilkinson notation provides a way to describe regression and repeated measures models without specifying coefficient values.