Oberseminarvortrag: Short Time-Series Prediction

am kommenden Montag, den 19.08.2019, hält Herr Son Pham, um 14:00 Uhr

den Oberseminarvortrag zu seinem Promotionsvorhaben.

Der Raum wird aufgrund des Urlaubs von Herrn Langer in Kürze mitgeteilt.

*****************************

Short Time-Series Prediction

ABSTRACT

Technical systems like vehicle or weapon systems, are typically highly complex and they are composed of a large number of parts and components.

To keep such systems highly available, it is important to predict how many of which spare parts are needed in the next years to have enough equipment to replace any component immediately when necessary.

In recent years, spare parts demand prediction has been extensively researched, but most models are not fully adequate due to the stochastic nature of the real environments. In many cases, it is required to predict the demand of hundred thousand spare parts, and this based only on the few data of their historical usage in the form of short time-series.

The main contribution of this work is the development of a novel framework able to handle the difficulties of short time-series prediction and a consistent number of spare parts. This framework uses a combination of unsupervised-learning and ensembles of supervised learners.

The first results prove the performance of this framework, not only in accuracy but also in terms of time-consumingefficiency.

****************************

Hierzu sind Sie herzlich eingeladen.