What-if forecast simulation: Statistical Forecasting with Damping Factors
In the previous What-if Wednesday previous posts, we experimented with the three parameters of the Holt-Winters method, namely; alpha, beta, and gamma. This week we are going to run some tests with the damping factor.
What is a Damping Factor?
Damping factor is an additional parameter that forecasting tools use to damp the forecast. This usually affects the trend of the forecast. Many a times we do not expect the sales to grow year over year with the same increasing (or decreasing) trend in which case we need to damp the forecast that was created using the historical trend. Its value can be set to any number between 0 and 1. The idea here is that when the damping factor is set to 0.9 the trend in the forecast will be reduced to 90% each future period.[Read Previous: What-if Wednesday: Seasonality Forecast Simulation (Gamma Parameter) ]
Damping Factor What-if Scenario Experiment
To experiment, we used a seasonal dataset that has a slight upward trend and applied the Holt-Winters forecasting method with varying damping factors. For our simulation, we have set the alpha, beta and gamma values to 0.05 and have varied the damping factor from 0.05 to 1 to understand its effect on the forecast. We could only see appreciable change in the forecast when the factor was 0.4 and above. So, the charts below show the forecast based on different damping factors as we go from 0.4 to 1. The actual forecast numbers are at the end of the post.
[Related: What-if Wednesday: Forecast Simulation – Changing The Beta Parameter in a Forecasting Method ]
The first chart where the damping factor is set to 0.4, we see the trend doesn’t really pick up much from its historical trend, meaning essentially no increase in the next two years. Now as we change the factor from 0.8 to 0.85 to 0.9 and so on we see an increase in the forecast as it follows the trend which is more in line with the history. And in the last chart, where the factor is 1, we see that the forecast continues with the same trend with no damping at all.
Ideally, we would not want to have a very high damping factor unless we are highly certain that the sales are going to continue to grow at this pace even far ahead into the future. We also do not want to keep it so low that the valuable trend information is lost. Typically, we keep it around 0.8 to 0.9 range so that it captures the trend along with a slight damping to be on the conservative side while forecasting.[Learn More: Statistical Demand Forecasting With Arkieva ]
When the historical data have a decreasing trend, this factor works in a similar fashion except that when the damping factor is high we would see a lower forecast. Basic idea stays the same that the trend decays as we go into the future.
Simulation Forecast Numbers
Want to join in our What-if-Wednesday posts? Add a comment below or send a tweet to @Arkieva with #WhatifWednesday, or email firstname.lastname@example.org to suggest a topic, scenario or simulation that you would like us to discuss.