fitclinear — Fit a binary linear classification model.

fitclinear fits a binary linear classifier from a predictor matrix or table and returns a ClassificationLinear object usable with predict.

Syntax

Mdl = fitclinear(X, Y)
Mdl = fitclinear(Tbl, ResponseVarName)
Mdl = fitclinear(Tbl, formula)
Mdl = fitclinear(Tbl, Y)
Mdl = fitclinear(___, Name, Value)
[Mdl, FitInfo] = fitclinear(___)

Inputs

Returns

Returned values from fitclinear depend on how many outputs the caller requests.

Errors

How fitclinear works

• fitclinear(X,Y) fits a binary SVM-style linear classifier with observations in rows of X and labels in Y.
• Table input supports fitclinear(Tbl,ResponseVarName), fitclinear(Tbl,formula), and fitclinear(Tbl,Y). Formula support covers additive predictor lists such as Y ~ A + B.
• Learner supports svm and logistic. Logistic models use logistic-loss training and return posterior-style scores by default; SVM models return raw signed-margin scores.
• Regularization supports ridge and lasso. Lambda accepts a nonnegative scalar or vector, producing one coefficient column per Lambda value.
• ObservationsIn supports rows and columns for numeric matrices. Table input requires row observations.
• Y can contain numeric, logical, string, char, or cell-string labels. The predicted label output preserves the label family.
• Name-value options ClassNames, PredictorNames, ResponseName, Weights, Prior, FitBias, Beta, Bias, Solver, IterationLimit, PassLimit, and ScoreTransform are accepted.
• Rows containing NaN in predictors or response, or zero weights, are omitted before fitting; Inf values and negative weights are rejected.
• Cross-validation, cost matrices, leaveout/holdout partitions, and hyperparameter optimization are not implemented yet and raise explicit errors instead of silently producing partial models.
• The returned object exposes Beta, Bias, Lambda, Learner, Regularization, Solver, ScoreTransform, ResponseName, PredictorNames, ClassNames, Prior, NumObservations, NumPredictors, and ModelParameters properties. [Mdl,FitInfo] returns training diagnostics including Lambda, Objective, NumIterations, NumPasses, and GradientNorm.

Examples

Fit and predict a logistic linear classifier

X = [0; 1; 2; 3];
Y = [0; 0; 1; 1];
[Mdl,FitInfo] = fitclinear(X, Y, 'Learner', 'logistic', 'Lambda', 0);
[label,score] = predict(Mdl, [0.2; 2.8])

Expected output:

label is [0; 1], and score contains one probability column per class.

Use column-oriented observations

Mdl = fitclinear([0 1 2 3], Y, 'ObservationsIn', 'columns');
label = predict(Mdl, [0.2 2.8], 'ObservationsIn', 'columns')

Expected output:

The predictor matrix is transposed internally so columns are observations.

Fit a table-input classifier

T = table(A, B, Y, 'VariableNames', {'A','B','Y'});
Mdl = fitclinear(T, 'Y ~ A + B', 'Learner', 'svm')

Expected output:

Mdl uses A and B as numeric predictors and Y as the binary response.

Using fitclinear with coding agents

Open a RunMat example with live inputs, then ask the agent to explain how fitclinear changes the result.

FAQ

Related Stats functions

Open-source implementation

Unlike proprietary runtimes, every RunMat function is open-source. Read exactly how fitclinear is executed, line by line, in Rust.

• View the source for fitclinear in Rust on GitHub
• Learn how the RunMat runtime works
• Found a bug? Open an issue with a minimal reproduction.

About RunMat

RunMat is an open-source runtime that executes MATLAB-syntax code blazing on any GPU. It is licensed under the Apache 2.0 license.

• RunMat automatically optimizes your math for GPU execution on Apple, Nvidia, and AMD hardware. No code changes needed. Simulations that took hours now take minutes.
• Start running code in seconds. RunMat runs in the browser, on the desktop, or from the CLI. No license server, no IT ticket.

Getting started · Benchmarks · Pricing