Exponential smoothing

[n_shinde2996]

What are Alpha, Beta, Gamma in Triple Exponential Smoothing?
What are the end results if we use this Smoothing?
How it is reflected in Demand Planning?

[lokesh_reddy4 · Accepted solution]

Hi Nikhil,

This is a more generic functional question and tried to give you simple answers to your questions.

Alpha, Beta, and Gamma are smoothing factors/parameters that can Exponential smoothing schemes weight past observations using exponentially decreasing weights to forecast future values. In simple words, recent data given more weight in forecasting than the older data.

The value of these smoothing factors ranges between 0 to 1. When the factor value is close to 1 then dampening(smoothed) older data is quick and when the factor value is close to 0 then the dampening of older data is slow.

The alpha value is used in a single exponential smoothing technique where there is no trend in past observations/data.

The single co-efficient Alpha is not enough when there is a Trend pattern in historical data and the need to introduce a new parameter Gamma in handling trends which are known as Double exponential smoothing. Always Gamma value must be chosen in conjunction with Alpha.

To handle seasonality along with trends then we need to introduce one more new parameter Beta. This seasonality some times called periodicity. This data (also known as hyperbola trends) is handled by the Triple exponential smoothing Technique.

Basically, these three techniques are intended to give more weight to recent data and less weight to data further in the past while generating the forecast for future periods.

Thanks and Regards,

Lokesh

[n_shinde2996 · Accepted solution]

Thanks for the update.