Privacy preserving collaborative forecasting based on dynamic exponential smoothing

The development and deployment of private forecasting technologies could allow supply chain collaborations to take place without revealing any participants´ data to the others, reaping the benefits of collaboration while avoiding the drawbacks. Atallah (2004) [1,3] is a step towards this goal, as it gives protocols for forecasting that reveal to the participants the desired answers yet do not reveal to any participant any other participant´s private data. But the smoothing coefficient OC used in Atallah (2004) [1,3] is assumed to be public and constant, but most of the time series, particularly in the complex economic system, many random observations of the sequence do not have a smoothing coefficient which does not change. Therefore, traditional exponential smoothing model for forecasting has a marked deviation, even serious distortion. So the assumption of constant is out of accordance with the practice. A novel part of this work is that the dynamic smoothing coefficient is established for exponential smoothing, and a corresponding privacy preserving collaborative forecasting algorithm is provided.