exponential
Exponential Smoothing Forecasting Method with Naïve start
Formula: Ft = α (At-1) + (1 – α) (Ft – 1) where:
Ft Forecast for time t
Ft – 1 Past forecast; 1 time ahead or earlier than time t At-1 Past Actual data; 1 time ahead or earlier than time t
α (read as alpha) as a smoothing constant takes the value between 0 and 1; or has value of either 0.1 to 0.9. With value more than 0 .5 (categorized as high), it gives more weight to recent data; with value less than 0.5 (categorized as low), it gives more weight to past data. With value equal to 0.5, forecast is giving emphasis on the past 3 periods; With value equal to 0.1, there is little weight on recent data and more on Past, about 19 periods in the past.
Naïve start is used in absence of forecast data. It is the actual prior period data to period t.
Sample Problem: Estimate the rate of increase in yield of mushroom for April and May in a given controlled field if the rate of increases in January and subsequent months are given below and other variables are same or constant:
Naïve Actual Data % Forecast January 10 - February 12 10 March 13
Formula: Ft = α (At-1) + (1 – α) (Ft – 1) where:
Ft Forecast for time t
Ft – 1 Past forecast; 1 time ahead or earlier than time t At-1 Past Actual data; 1 time ahead or earlier than time t
α (read as alpha) as a smoothing constant takes the value between 0 and 1; or has value of either 0.1 to 0.9. With value more than 0 .5 (categorized as high), it gives more weight to recent data; with value less than 0.5 (categorized as low), it gives more weight to past data. With value equal to 0.5, forecast is giving emphasis on the past 3 periods; With value equal to 0.1, there is little weight on recent data and more on Past, about 19 periods in the past.
Naïve start is used in absence of forecast data. It is the actual prior period data to period t.
Sample Problem: Estimate the rate of increase in yield of mushroom for April and May in a given controlled field if the rate of increases in January and subsequent months are given below and other variables are same or constant:
Naïve Actual Data % Forecast January 10 - February 12 10 March 13