Analysis of Uncertainty in Time Series Data: Issues and Challenges

Abstract

This paper reviews issues and challenges of uncertainty in time series data. The aim of uncertainty analysis is to determine the ways of how to deal with uncertain data in order to gain knowledge, fit low dimensional model, and do prediction. So as to build an efficient predictive tool, uncertainty in data could not be ruled out because it may bring important knowledge. Uncertainty information arises from different resources such as process uncertainty, model uncertainty or data uncertainty. In this paper, issues and challenges of these uncertainties in time series data will be disscussed and how data mining methods could be used to solve it. Frequent pattern mining algorithm through FP-growth, Apriori algorithm and H-mine are methods that could be used to investigate the existing of uncertainty data. While, Euclidean distance, particle swarm optimization, Monte Carlo simulation, and regression are methods that could be compared to investigate the best ways of predicting These methods have been implemented in many types of data since early 1900s. In this paper, the characteristic of uncertainty in time series data will be discussed and uncertainty tests are conducted as to prove the existing of uncertainty in the selected time series data. This work will benefit in many application domains especially in the weather domain.



Author Information
Nabilah Filzah Mohd. Radzuan, University Kebangsaan Malaysia, Malaysia
Zalinda Othman, University Kebangsaan Malaysia, Malaysia
Azuraliza Abu Bakar, University Kebangsaan Malaysia, Malaysia

Paper Information
Conference: ACTIS2014
Stream: New Realities through Artificial Intelligence

This paper is part of the ACTIS2014 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window

Posted by amp21