How to forecast a multi step ahead time series?

Hello all,

I'm doing a new project on MATLAB. I got stuck in a step where I have to forecast a multi step ahead time series.

Any code snippet will be very helpful for me to proceed. Kindly help.

Problem :

I have a time series data of 578 points and I need to forcast next 144 points which are unknown to me. These 144 points will be multi step ahead forecast points.

The problem can be seen as I have 80 percent data with me and rest unknown 20 percent I need to predict. (Any relevant approach is great for me)

Thanks in advanced.

Answer 1

[Jaynik]

Hi Atreyee,

The predict function from the "System Identificatoin Toolbox" can be used to perform multi-step ahead forecasting. This function computes the K-step-ahead output of an identified model using measured input-output data. Here is a sample code for the same:

% Assuming 'data' is your time series data
% Estimate an AR model from the data
Mdl = ar(data, 2);

% Perform the K-step ahead prediction
K = 144;
yp = predict(Mdl, data, K);

% Plotting the original data and the prediction
figure
plot(data.OutputData)
hold on
plot(yp.OutputData, 'r')
hold off

Here, ar is a model used to estimate an autoregressive model of order 2 from the data. The choice of model depends on the characteristics of your data and the specific requirements of your forecasting task. It is always a good idea to try multiple models and choose the one that gives the best performance on your validation data.

Refer to the following documentation to learn more about these functions:

• predict: https://www.mathworks.com/help/ident/ref/predict.html
• ar: https://www.mathworks.com/help/ident/ref/ar.html

Here are some examples for time-series prediction which might be useful:

• https://www.mathworks.com/help/deeplearning/gs/get-started-with-time-series-forecasting.html
• https://www.mathworks.com/help/deeplearning/ug/time-series-forecasting-using-deep-learning.html
• https://www.mathworks.com/help/deeplearning/ug/time-series-forecasting-using-deep-network-designer.html

Hope this helps!