run

Execute pipeline or component for inference after learning

Since R2026a

Syntax

output1 = run(learnedPipe,input1,...,inputN)

[output1,...,outputM] = run(learnedPipe,input1,...,inputN)

___ = run(___,Name,Value)

Description

output1 = run(learnedPipe,input1,...,inputN) uses the provided inputs to evaluate all nodes in the pipeline or component learnedPipe that are required to return output1. The function ignores nodes that are not required to return the output.

[output1,...,outputM] = run(learnedPipe,input1,...,inputN) also returns the outputs corresponding to the pipeline or component outputs in the Outputs property. You can see the identity and order of the pipeline outputs using learnedPipe.Outputs.

___ = run(___,Name,Value) specifies optional run parameters in addition to any of the input or output argument combinations in previous syntaxes. The software forwards the run parameter values to individual components during the execution of learnedPipe.

• If learnedPipe is a LearningPipeline object, then specify Name as the component name with / as a prefix to the parameter name, to disambiguate between components that have the same parameter name. For example, specify run(learnedPipe,X,Y,"ClassificationSVM/LossFun","classifcost").

• If learnedPipe is a learning component object, then specify Name as the parameter name. In this case, you can use the Name=Value syntax. For example, specify run(learnedPipe,X,Y,LossFun="classifcost"). Use help(learnedPipe) to see the available run parameters.

Examples

Learn the Pipeline Parameters and Predict Responses on New Data

Create a pipeline with two components: one that performs principal component analysis and retains three components, and one that performs ECOC classification.

pca = pcaComponent(NumComponents=3);
ecoc = classificationECOCComponent;
pipeline = series(pca,ecoc);

Load sample data and partition the data into training and test data.

fisheriris = readtable("fisheriris.csv");
rng("default")
cvp = cvpartition(fisheriris.Species,Holdout=0.25,Stratify=true);

idxTrain = training(cvp);
Xtrain = fisheriris(idxTrain,1:end-1);
Ytrain = fisheriris(idxTrain,end);

idxTest = test(cvp);
Xtest = fisheriris(idxTest,1:end-1);
Ytest = fisheriris(idxTest,end);

Evaluate the pipeline using the training data.

learnedPipeline = learn(pipeline,Xtrain,Ytrain);

Compute the predicted response values for the test data, and plot the confusion matrix to compare the results with the actual labels. Provide only the input required for predicting the response values.

Ypredicted = run(learnedPipeline,Xtest);
confusionchart(Ytest{:,:},Ypredicted{:,:})

You can ask for additional outputs in the order specified by the pipeline Outputs property. Compute the predictions, classification scores, and loss on the test set.

[predictions,scores,loss] = run(learnedPipeline,Xtest,Ytest);

Run Pipeline for Different Numbers of Principal Components

Create a pipeline that performs principal component analysis and classification using ECOC.

pca = pcaComponent;
ecoc = classificationECOCComponent;
pipeline = series(pca,ecoc);

Load sample data and define the predictor and response variables.

fisheriris = readtable("fisheriris.csv");
X = fisheriris(:,1:end-1);
Y = fisheriris(:,end);

Execute the pipeline for different numbers of principal components, and plot the loss values.

numComponents = 1:4;
losses = zeros(1,numel(numComponents));

for j = 1:numel(numComponents)
    pipeline.Components.PCA.NumComponents = numComponents(j);
    learnedPipeline = learn(pipeline,X,Y);
    [~,~,losses(j)] = run(learnedPipeline,X,Y);
end

plot(numComponents,losses,"o:")
xlabel("Number of Components")
ylabel("Loss")

The resubstitution loss value decreases as the number of principal components increases.

Input Arguments

learnedPipe — Pipeline or component to evaluate

LearningPipeline object | learning component object

Pipeline or component to evaluate, specified as a LearningPipeline object or a learning component object in one of the following tables. Any data-dependent parameters (learnables) in learnedPipe must already be learned.

Data Preprocessing Components

Component	Purpose
equalWidthBinnerComponent	Grouping data into equal-width bins
frequencyEncoderComponent	Frequency encoding categorical variables
kmeansEncoderComponent	Feature extraction using k-means clustering
normalizerComponent	Normalizing data
observationImputerComponent	Imputing missing values
observationRemoverComponent	Removing observations
oneHotEncoderComponent	Encoding categorical data into one-hot vectors
outlierImputerComponent	Imputing outlier values
outlierRemoverComponent	Removing outlier values
pcaComponent	Principal component analysis (PCA)
quantileBinnerComponent	Binning data into equally probable bins
ricaComponent	Feature extraction using reconstruction independent component analysis (RICA)
sparseFilterComponent	Feature extraction using sparse filtering

Feature Selection and Engineering Components

Component	Purpose
featureSelectionClassificationANOVAComponent	Feature selection using one-way ANOVA test
featureSelectionClassificationChi2Component	Feature selection using chi-square tests
featureSelectionClassificationKruskalWallisComponent	Feature selection using Kruskal-Wallis test
featureSelectionClassificationMRMRComponent	Minimum redundancy maximum relevance (MRMR) feature selection in classification workflow
featureSelectionClassificationNCAComponent	Neighborhood component analysis (NCA) feature selection in classification workflow
featureSelectionClassificationReliefFComponent	ReliefF feature selection in classification workflow
featureSelectionRegressionFTestComponent	Feature selection using F-tests
featureSelectionRegressionMRMRComponent	Minimum redundancy maximum relevance (MRMR) feature selection in regression workflow
featureSelectionRegressionNCAComponent	Neighborhood component analysis (NCA) feature selection in regression workflow
featureSelectionRegressionReliefFComponent	ReliefF feature selection in regression workflow
variableSelectorComponent	Manual variable selection

Classification Model Components

Component	Purpose
classificationDiscriminantComponent	Discriminant analysis classification
classificationECOCComponent	Multiclass classification using error-correcting output codes (ECOC) model
classificationEnsembleComponent	Ensemble classification
classificationGAMComponent	Binary classification using generalized additive model (GAM)
classificationKernelComponent	Classification using Gaussian kernel with random feature expansion
classificationKNNComponent	Classification using k-nearest neighbor model
classificationLinearComponent	Binary classification of high-dimensional data using a linear model
classificationNaiveBayesComponent	Multiclass classification using a naive Bayes model
classificationNeuralNetworkComponent	Classification using a neural network model
classificationSVMComponent	One-class and binary classification using a support vector machine (SVM) classifier
classificationTreeComponent	Decision tree classifier

Regression Model Components

Component	Purpose
regressionEnsembleComponent	Ensemble regression
regressionGAMComponent	Regression using generalized additive model (GAM)
regressionGPComponent	Gaussian process regression
regressionKernelComponent	Kernel regression using explicit feature expansion
regressionLinearComponent	Linear regression
regressionNeuralNetworkComponent	Neural network regression
regressionSVMComponent	Regression using a support vector machine (SVM)
regressionTreeComponent	Decision tree regression

Custom Components

Component	Purpose
functionComponent	Custom function

input — Input data required by pipeline or component

table

Input data required by the pipeline or component learnedPipe, specified as a table. Input data can be predictor data, response values, observation weights, and so on. The order of the inputs 1, …, N must match the order of the pipeline or component inputs, as listed in the Inputs property. You can see the identity and order of the pipeline inputs using learnedPipe.Inputs.

Data Types: table

Output Arguments

output — Output data

separate variables

Output data computed by the component based on the input data, returned as separate variables.

Version History

Introduced in R2026a