forecasting in regression learner

hello everybody

is there a way to forecast n steps ahead the results of a regression learner on a time series (predicted values), choosing by toolstripes and avoiding formulas in command window? tks

(edit: if not please give me a simple formula for forecasting n steps ahead a model saved in workspace)

Answer 1

[Hornett]

Hi roberto,

You will need to use the "predict" function in a loop to forecast steps in regression learner. Below is the example code for the same.

function forecastedValues = forecastNStepsAhead(trainedModel, initialData, nSteps)
% trainedModel: The exported regression model
% initialData: The initial data to start forecasting from
% nSteps: Number of steps to forecast ahead

forecastedValues = zeros(nSteps, size(initialData, 2));
currentData = initialData(end, :); % Start with the last available data point

for i = 1:nSteps
    % Predict the next value
    predictedValue = predict(trainedModel, currentData);
    
    % Store the predicted value
    forecastedValues(i, :) = predictedValue;
    
    % Update the current data for the next prediction
    currentData = predictedValue;
end
end

% Example usage
initialData = data(end, :); % Use the last row of your original data as the starting point
nSteps = 10; % Number of steps to forecast ahead
forecastedValues = forecastNStepsAhead(trainedModel, initialData, nSteps);

% Display the forecasted values
disp('Forecasted Values:');
disp(forecastedValues);

• predict: https://www.mathworks.com/help/stats/linearmodel.predict.html

Hope this helps!