. The novel Time Series Data Mining (TSDM) framework is applied to analyzing financial time series. The TSDM framework adapts and innovates data mining concepts to analyzing time series data. In particular, it creates a set of methods that reveal hidden temporal patterns that are characteristic and predictive of time series events. This contrasts with other time series analysis techniques, which typically characterize and predict all observations. The TSDM framework and concepts are reviewed, and the applicable TSDM method is discussed. Finally, the TSDM method is applied to time series generated by a basket of financial securities. The results show that statistically significant temporal patterns that are both characteristic and predictive of events in financial time series can be identified. 1 Introduction The Time Series Data Mining (TSDM) framework [1-4] is applied to the prediction of financial time series. TSDM-based methods can successfully characterize and p...

In this paper, we describe the process of discovering underlying knowledge in a set of isokinetic tests, using a new algorithm to find similar patterns in a set of temporal series. An isokinetic machine is basically a physical support on which patients exercise one of their joints, in this case the knee, according to different ranges of movement and at a constant speed. The data on muscle strength supplied by the machine are processed by an expert system that has built-in knowledge elicited from an expert in isokinetics. It cleans and pre-processes the data and conducts an intelligent analysis of the parameters and morphology of the isokinetic curves. Then, Data Mining methods based on the discovery of sequential patterns in time series by means of which to find similarities and differences among exercises were applied to the processed information to characterise injuries of those patients. The results obtained were applied in two environments: one for the blind and another for elite athletes.

A new framework for analyzing time series data called Time Series Data Mining (TSDM) is introduced. This framework adapts and innovates data mining concepts to analyzing time series data. In particular, it creates methods that reveal hidden temporal patterns that are characteristic and predictive of time series events. The TSDM framework, concepts, and methods, which use a genetic algorithm to search for optimal temporal patterns, are explained and the results are applied to real-world time series from the engineering and financial domains.

A comparison of three methods for saving previously calculated fitness values across generations of a genetic algorithm is made. These methods lead to significant computational performance improvements. For real world problems, the computational effort spent on evaluating the fitness function far exceeds that of the genetic operators. As the population evolves, diversity usually diminishes. This causes the same chromosomes to be frequently reevaluated. By using appropriate data structures to store the evaluated fitness values of chromosomes, significant performance improvements are realized. Several different data structures are

Most of the existing data mining approaches to time series prediction use data preparation techniques involving an embed of the most recent values of the time series, following the traditional linear auto-regressive methodologies. However, in many time series prediction tasks the alternative approach that uses derivative features constructed from the raw data with the help of domain theories can produce significant prediction improvements. This is particularly noticeable when the available data includes multivariate information but the aim is still the prediction of one particular time series, a situation that occurs frequently in financial time series prediction. This paper presents a method of feature construction based on domain knowledge that uses multivariate time series information and improves the accuracy of next-day stock quotes prediction, when compared with the traditional embed of historical values extracted from the original data.

In this paper, the dual track use of the time stepping coupled finite element-state space modeling of induction motors to generate databases for healthy and faulty motor performances, coupled to time series data mining techniques, is presented. This dual track is demonstrated here in its embryonic stage to represent a potentially very powerful motor fault diagnostics and identification tool, when fully developed to completion. Thus, the paper presents results that point to a potentially very useful technical tool for fault diagnostics and preventative maintenance of electric motor-drive systems. I. Introduction Three-phase induction motors are the machine of choice in the majority of electronically controlled adjustable/variable speed drive (ASD) applications. In the not too distant future, a widening use of induction motors/ASDs for naval vessel propulsion is expected [1]. An important area of study during the past twenty years is the analysis and diagnosis of in...

Autoregressive integrated moving average (ARIMA) models are used in different researches for modelling and forecasting of traffic and Quality of Service (QoS) parameter values in telecommunication networks to make reasonable short, medium- and long-term predictions. We propose methodology to use ARIMA models for QoS prediction in network scenarios based on a preliminary detection and elimination of outliers in the time series data. Outliers are patterns describing QoS behaviour caused by fault events (route failure, operation anomalies, DoS attacks, misconfiguration, etc). Practically, we show the feasibility of combination of ARIMA prediction with outlier detection for short and medium term forecasting (daily and monthly) using real world end-to-end delay QoS measurement data.

A new method for analyzing time series data is introduced in this paper. Inspired by data mining, the new method employs time-delayed embedding and identifies temporal patterns in the resulting phase spaces. An optimization method is applied to search the phase spaces for optimal heterogeneous temporal pattern clusters that reveal hidden temporal patterns, which are characteristic and predictive of time series events. The fundemantal concepts and framework of the method are explained in detail. The method is then applied to the characterization and prediction, with a high degree of accuracy, of the release of metal droplets from a welder. The results of the method are compared to those from a Time Delay Neural Network and the C4.5 decision tree algorithm.

This paper presents the results from characterizing and predicting the release of droplets of metal from a welder. The welding process joins two pieces of metal into one by making a joint between them. An arcing current melts the tip of a wire, forming a metal droplet that elongates until it releases. The goal is to predict the moment when a droplet will release, which can improve the quality of the joint by allowing the droplet releases to be monitored and controlled.

Abstract. Many business and scientific domains require the collection and analysis of sequences of events and time series data. Although statistical approaches have been long applied to time series, most of these approaches assume the time series is stationary and typically must be applied globally to the sequence. Thus, other methods are needed to solve many types of problems that occur in sequential business and scientific applications. One such problem is when there is no global correlation between sequences, but there are periodic occurrences when the signature of one sequence is present in other sequences. This can be solved by using association rules that relate the sequences of events, as described in this dissertation.