XLMiner facilitates the analysis of datasets via the use of trend discovery techniques (autocorrelation and partial autocorrelation) and comprehensive modeling methods (ARIMA and exponential smoothing).
The XLMiner solution enables users to select from a variety of smoothing techniques, from classic exponential smoothing (weighted averages of historical observations with weights decaying exponentially as the observations age) to the more advanced Holt-Winters technique (datasets featuring both trends and seasonaliy). XLMiner supports four different tools used to reduce the effect of random data variations (aka smoothing): Exponential, Double Exponential, Moving Average, and Holt-Winters.
Smoothing Models in XLMiner
There are multiple smoothing methods available in XLMiner, including:
- Exponential: Assignation of exponentially decreasing weights starting with the most recent observations. The forecast will be a constant value which is the smoothed value of the last observation. Do not use this model when seasonality is present.
- Moving Average: In this technique, each observation is assigned an equal weight. Additional observations are forecasted by using the average of the previous observations. Do not use this model when seasonality is present.
- Double Exponential: Similar to exponential smoothing, except an additional calculation with a trend parameter is introduced. This technique should be used when a trend is apparent in the dataset but, like exponential smoothing, should not be used when seasonality is present.
- Holt-Winters: When data conveys both a trend as well as seasonality, then the Holt-Winters method is the most appropriate smoothing technique. In these cases, a third parameter is introduced to account for seasonality (periodicity) in a dataset. The Holt-Winters equation set are ideal for datasets featuring both trends and seasonality.
XLMiner is capable of implementing any of the above smoothing techniques in order to achieve accurate time series forecasting.
How to Access Smoothing Settings in Excel
- Launch Excel.
- In the toolbar, click XLMINER PLATFORM.
- In the ribbon, click Smoothing.
- In the drop-down menu, select either Exponential, Double Exponential, Moving Average, or Holt-Winters. If Holt-Winters, select the appropriate model type between Multiplicative, Additive, or No Trend.
Smoothing Model Summary
- Used to highlight relevant data and correct autocorrelated errors.
- XLMiner features four smoothing models for selection, the choice of which is typically based on specific trend/seasonality requirements.
- Time Series Example: View an example of how an ARIMA model can be applied.
- Using Time Series: How to use time series analysis functionality within XLMiner.
- ARIMA Models: How XLMiner uses ARIMA modeling to fit datasets.
- XLMiner Online Help: Help system covering functionality within the XLMiner module.