# Statistical Forecasting using Forecast Smoothing

Sales and Operation Planning tool ( S&OP) uses Single Exponent , Double Exponent and Triple Exponent smoothing to predict the values in future for demand planning. These smoothing or time series analysis algorithms use ALPHA, BETA AND GAMMA factors to take into consideration the historical data , trending data or any seasonality trends.

More information on these Exponent smoothing methods is available here on Predictive Analysis on HANA Cloud Platform and Sales & Operations Planning ( S&OP) Statistical Forecasting .

The predicted  results are based on Alpha, Beta & Gamma factors value. To predict future data by find the best fitting data is mostly trial and error method which can be quite tedious and may involve lot of effort to determine the right or best values of these factors , so that predicted value fit with the historical data.

S&OP eases on the effort part by maintaining multiple values for Alpha, Beta & Gamma factors in its settings so when the statistical forecast process is executed it runs with each set of values and automatically picks up results of the best or for which result set is in least statistical error  between the current/historical and predicted data.

With HANA SPS 07 (&08) a new time series analysis algorithm is available called Forecast smoothing . Forecast Smoothing is used to calculate optimal parameters of a set of smoothing functions,  including Single Exponential Smoothing, Double Exponential Smoothing, and Triple Exponential Smoothing.

Different set of Alpha , Beta and Gamma factors need not be maintained. It automatically runs the statistical forecasting based on the statistically optimal algorithm and parameters. It does quality assessment comparing historic and forecast values and automatically chooses the one with least error. Forecasting result are not limited to set of manually maintained values .

We can automatically define or estimate parameters of what’s the best type or appropriate smoothing ( single/ double or smoothing )  using forecast smoothing.

For example , for the same data used previously with manually given parameters to test and best fit single, double and triple exponent smoothing , the  forecast smoothing technique automatically analysed and came up with Triple exponent smoothing with Alpha = 0.127 , Beta = 0.0000001 and Gamma  = 0.883 as the best fit, & these kind of values are unlikely to be manually determined.

Find below images of results from previous examples to compare with forecast smoothing, which easily fits better .