What is backcasting wit reference to setting initial conditions?

Backcasting is an option when setting initial conditions for model fitting, specifically "Specifying Initial Conditions for Iterative Estimation of Transfer Functions". Documentation says "Estimates initial conditions using a backward filtering method (least-squares fit)." which is a little vague. What is backcasting really?

Answer 1

[Sai Pavan]

Hi Damon,

I understand that you are trying to learn more about backcasting method that is available as an option when setting initial conditions in transfer function model fitting.

Backcasting is a method used to specify initial conditions for iterative estimation of transfer functions that is typically performed iteratively.

• It involves running the model in reverse order, starting from the last data point and working backward to the first data point.
• Initially, an initial guess for the model's parameters and initial conditions is used.
• The model is then fitted to the data using backcasting and a least-squares fit. This means that the model's parameters are adjusted to minimize the difference between the model's output and the observed data.
• The estimated initial conditions are then used as the new initial conditions for the next iteration, and the process is repeated until convergence is achieved. This process helps to obtain better estimates of the initial conditions for accurate model fitting and estimation.

Hope this information will help you understand the backcasting method better.

Regards,

Sai Pavan