Share Email Print

Proceedings Paper

Data mining on time series of sequential patterns
Author(s): Ari J. E. Visa
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

Ordinary Time Series Analysis has long tradition in statistics [3] and it has also been considered in Data Mining [4,13]. Sequential patterns that are common in many measurements in process industry and in elsewhere have also been considered [1,14,11] in Data Mining. However, in some cases these two approaches can be merged together into a suitable transform. This kind of 2D-transform should be selected such a way that the basis functions support the Data Mining and the interpretation of results. As an example a runnability problem on a paper machine was considered. There were problems with fluctuations in paper basis weight. Data mining was successfully applied to the problem to identify and to remove the disturbances. The whole disturbance analysis was based on 86 sequential patterns consisting of 62 point-wise measurements in cross direction. These measurements were acquired from the process control system. The consecutive 86 patterns were Slant-transformed and the results were data mined. It was quite easy to find out the uneven static distribution of pressure in the head box and to find that the pressure fluctuated in the head box. Based on the considered case it can be claimed that Data Mining might be a good tool in many trouble shooting problems.

Paper Details

Date Published: 12 March 2002
PDF: 6 pages
Proc. SPIE 4730, Data Mining and Knowledge Discovery: Theory, Tools, and Technology IV, (12 March 2002); doi: 10.1117/12.460225
Show Author Affiliations
Ari J. E. Visa, Tampere Univ. of Technology (Finland)

Published in SPIE Proceedings Vol. 4730:
Data Mining and Knowledge Discovery: Theory, Tools, and Technology IV
Belur V. Dasarathy, Editor(s)

© SPIE. Terms of Use
Back to Top