Multiple Input Multiple Output Gaussian Regression Model

Is there any Multiple Input Multiple Output Gaussian Regression Model function in matlab?

Or, should i estasblish it myself?

Answer 1

[Ayush]

Hi @hanqiu bao,

Yes, MATLAB provides a function called fitgmdist that can be used to create a Multiple Input Multiple Output (MIMO) Gaussian regression model. This function is part of the Statistics and Machine Learning Toolbox.

Here's an example of how you can use fitgmdist to create a MIMO Gaussian regression model:

% Generate random data for demonstration
X = randn(100, 2);  % Input variables
Y = X(:, 1) + 2*X(:, 2) + randn(100, 1);  % Output variables

% Fit MIMO Gaussian regression model
numComponents = 2;  % Specify the desired number of Gaussian components as an integer
model = fitgmdist(X, numComponents);

% Predict using the trained model
X_new = randn(10, 2);  % New input data for prediction
Y_pred = posterior(model, X_new);  % Predict the output variables

% Display the predicted output
disp(Y_pred);

    0.9711    0.0289
    0.9909    0.0091
    0.9673    0.0327
    0.9821    0.0179
    0.4084    0.5916
    0.4268    0.5732
    0.9999    0.0001
    0.5254    0.4746
    0.0519    0.9481
    0.7809    0.2191

Hope it helps!

[hanqiu bao]

Hi, @Ayush

Thanks for your help!

As far as I know, fitgmdist seems to be a clustering method. :)

In matlab, there has a function fitrgp, but this is a miso function.

Indeed, I want to find a mimo function like fitrgp.

[Ayush]

You're welcome! Yes, you are correct that fitgmdist is a clustering method in MATLAB. It is used to fit a Gaussian mixture model to a dataset.

Regarding your question, if you are looking for a MIMO (Multiple Input Multiple Output) regression function similar to fitrgp in MATLAB, you can consider using the fitrsvm function. fitrsvm is used for training Support Vector Machine (SVM) regression models, which can handle multiple input and output variables.

Read more about it here.