You make a manual model selection in order to determine a specific forecast strategy. This is appropriate if the pattern generated with the automatic model does not lead to the desired result.

If you want to select a model manually, you must first analyze the historical data to determine whether a distinct pattern or trend exists. You then define your forecast model accordingly.

Here, we describe how you can react to different patterns using the forecast's features:

If your historical data represents a constant pattern, you can select either the constant model or the constant model with adaptation of the smoothing factors. In both cases, the forecast is carried out using first-order exponential smoothing. When adapting the smoothing parameters, the system calculates different parameter combinations and then selects the optimum parameter combination. The optimum parameter combination is the one which results in the lowest mean absolute deviation.

You have another two options if the historical pattern is constant; either the moving average model or the weighted moving average model.

If your historical data represents a trend, you should select either the trend model or a second-order exponential smoothing model. In the trend model, the system calculates the forecast values using first-order exponential smoothing.

In the second-order exponential smoothing models, you can choose a model with or without parameter optimization.

If your historical data represents a seasonal pattern, you can specify the seasonal model. The system calculates the forecast values for the seasonal model using first-order exponential smoothing.

If your historical data represents a seasonal trend pattern, you can select a seasonal trend model. The system calculates the forecast values using first-order exponential smoothing.

If you cannot detect any of the above trends or patterns, and you still want the system to carry out a forecast, it is advisable to select either the moving average model or the weighted moving average model.