Choice of moving average or exponential smoothing for a particular product profile
Introduction
The demand data for a product has been shown in the table below. Compare the forecasts using a Moving Average with a period of 5 months, MA(5), and an Exponential smoothing Method with an α of 0.33. For Exponential Smoothing use the midpoint of first 5 month range of the average as the initial Forecast. (Hint: the Exponential Smoothing Forecast will be initialized with a forecast of 4951 for April made in March.)
Visualization of raw data
Forecasting with 5-point moving average & simple exponential smoothing
Here we use α=1/3 for our exponential smoothing model
Accuracy of the models
To estimate the accuracy of the models, we first compare the Mean Absolute Deviation (MAD) for each of the models.
5MA | SES | |
---|---|---|
MAD | 453.7429 | 432.0553 |
Also comparing the error cumulatively gives a better picture of the accuracy of each of the models.
Conclusion
For this particular product, the forecasts from June to December show that SES is performing better than 5MA. SES outperforms 5MA for 5 months while 5MA outperforms SES for only 2 months.