Discovery of product demand anomalies by machine learning algorithms

OData support
Supervisor:
Dr. Szűcs Gábor
Department of Telecommunications and Media Informatics

Nowadays, machine learning is increasingly present, and business can not be excluded from that. The reason for gaining ground lies in usability, because machine learning can be used in many different areas with great efficiency.

In my degree, I use machine learning to track changes in time series, which includes the trend and level changes and the detection of anomalies.

In this thesis, the goal is to create a system that can detect anomalies in the moment of their emergence. During the construction of such a system, the understanding and preparation of the data play an important role, and so I put great emphasis on it.

In my dissertation, I work on real business data, which comes from a production system of one of the biggest SAP's partners, and time series contain order data.

I have been handling the task to be performed as a classification problem, hence I also have to work out an automatic annotation procedure. To solve the classification task I used the widely used SVM classification algorithm.

Downloads

Please sign in to download the files of this thesis.