It is recommended to have at least 2–3 full seasonal cycles in the historical data. For this to work properly, the more repeating cycles the historical data contains the better. This forecasting method looks for seasonality patterns in the historical data and tries to determine the pattern that best matches the data. In some cases we know what the seasonality length is, but in other cases we do not. Since the data is monthly and repeats every 12 points, the detected seasonality was 12. In the example below you can see how a yearly seasonality was detected and applied in the forecast. We would expect to have a yearly cycle in this case, which would repeat itself every 12 points (months). Another example is hourly traffic data, where a seasonality of 24 points (hours) makes sense. An example of this could be ice cream sales presented in monthly data.

In many business scenarios there is a seasonality pattern that we would like to take into account in the forecast. The main advantages of using the ETS method are the ability to detect seasonality patterns and confidence intervals. Exponential Smoothing methods are a popular way to forecast and are among the leading methods that have become industry standards. The new functionality in Excel 2016 utilizes another algorithm, called Exponential Smoothing or ETS. Before Excel 2016, many used the FORECAST() sheet function, which performs a linear forecast or extended trendlines in chart properties to extrapolate forward. There are many ways to generate a forecast for your historical data.

Get Excel Using Exponential Smoothing for forecasting

Forecast function in excel for mac