Prediction of obsolescence degree as a function of time: A mathematical formulation

Predicting the obsolescence risk of components is an important challenge in system life-cycle management to improve its durability. The obsolescence degree measures the obsolescence occurrence risk. Thus, it expresses the probability that an entity will become obsolete within a given time horizon. Some mathematical techniques have already been proposed to deal with this problem. Recent studies have also shown that machine learning methods can be an effective way to improve prediction capability. However, these two classes of techniques assess the obsolescence degree by a scalar at observation time. This does not permit the projection of its possible evolution over time. This paper proposes a new approach based on probability distribution to model the obsolescence degree as a function of time. Based on sales data, the obsolescence degree is modeled as a function of time, and the remaining time-to-obsolescence is inferred. The proposed approach is tested in the prediction of smartphones’ obsolescence. This study's results are then analyzed while proposing future work to be carried out to increase the approach's applicability.