The bootstrp function in MATLAB is used to perform bootstrapping, which involves resampling your data with replacement and computes a statistic by applying a function to each bootstrap sample. The fitlm function is not directly compatible with bootstrp because fitlm expects data in a specific format and returns a model object, not a simple numeric output that bootstrp can handle.
To use bootstrp with fitlm, you need to define a custom function that fits the model and returns a numeric statistic for instance the coefficients of a linear regression model as shown below:
% Define a custom function that fits a linear model and returns coefficients
function coefficients = fitlm_coefficients(data)
X = data(:, 1:end-1); % All columns except the last one
y = data(:, end); % The last column
mdl = fitlm(X, y, 'linear');
coefficients = mdl.Coefficients.Estimate'; % Return coefficients as a row vector
end
% Combine X and y into a single matrix for bootstrapping
data = [X, y];
% Perform bootstrapping
bootCoefficients = bootstrp(500, @fitlm_coefficients, data);
This approach allows you to bootstrap the coefficients or any other statistic derived from the linear model.
