standard error of regression from fitlm

Hi Everyone:

I am using fitlm function for linear regression of some data. How do I find the standard error of the regression? I do not want the standard error of the coefficients. Please help.

Answer 1

[Gautham Sholingar]

The linear model created by using the fitlm command has properties like MSE, Rsquared and SSE (Sum of Squared Errors) which should give you the data you want.

    load hald

    linearModel = fitlm(ingredients,heat);
    % Mean Squared error
    linearModel.MSE
    linearModel.Rsquared
    linearModel.SSE % Sum of squared errors

    % Predict using a different data set
    % using the same data set here for demonstration
    prd = predict(linearModel,ingredients)

    % compare to original data and measure mean squared 
    heat % replace with expected values

    % sum of squared errors
    error = sum((prd - heat).^2)

In addition, you can use the linear model to predict the output for a different data set and then use the method shown in the above code to compute the sum of the squared errors.