Lecture 6 Exponential Smoothing

Exponential smoothing methods are more suitable for:

• Long term forecasts

• Short term forecasts

• Short and medium term forecasts (it depends)

• Only Medium term forecast

In Simple exponential Smoothing (SES), for data with high noise, proposed values for level smoothing parameter are:

• a -> 0
• a -> 1
• a -> 0.5
• a -> 2 

In case of Simple exponential Smoothing (SES), if level smoothing parameter is equal to 0 then forecasts of Simple Exponential Smoothing (SES) are the same with:

• Theta

• naïve

• moving average

• first forecast

In Simple exponential Smoothing (SES), for data with no constant average value, proposed values for level smoothing parameter are:

• a -> 0
• a -> 1
• a -> 0.5
• a -> 2

Linear Exponential Smoothing (Holt) has been accused as optimist because:

• Mean error is always positive

• Mean error is always negative

• Mean error is always zero

Value of trend smoothing parameter (b) must be higher than value of level smoothing parameter (a)

• True
• False

Which is the proposed method to find the value smoothing parameters in exponential smoothing

• Minimize mean squared error
• Μaximize mean squared error
• Depicting seasonality
• Decomposition

In case of Simple exponential Smoothing (SES), if level smoothing parameter is equal to 1 then Simple Exponential Smoothing (SES) is the same with:

• Holt

• Damped

• Naive

• Moving Average