Today, real world time series data sets can take a size up to a trillion observations and even more. Data miners’ task is it to detect new information that is hidden in this massive amount of data. While well known techniques for data mining in cross sections have been developed, time series data mining methods are not as sophisticated and established yet. Large time series bring along problems like very high dimensionality and up to today, researchers haven’t agreed on best practices in this regard. This review gives an overview of the challenges of large time series and the proposed problem solving approaches from time series data mining community. We illustrate the most important techniques with Google trends data. Moreover, we review current research directions and point out open research questions. Heutzutage sind die Möglichkeiten der Datensammlung und -Speicherung unvorstellbar weitreichend und somit können Zeitreihendatensätze mittlerweile bis zu einer Billion Beobachtungen enthalten. Die Aufgabe von Data Mining ist es, versteckte Informationen aus dieser Datenschwemme herauszufiltern. Während es für Querschnittsdaten viele verschiedene und sehr gut entwickelte Techniken gibt, hinken die Zeitreihen Data Mining Methoden weit hinterher. Die Forschungspraxis hat sich in diesem Bereich noch nicht auf standardisierte Vorgehensweisen geeinigt. Dieser Literaturüberblick stellt zunächst die typischen Probleme, die Zeitreihen mit sich bringen, dar und systematisiert daraufhin die von der Forschungsgemeinde vorgeschlagenen Lösungsansätze hierfür. Die wichtigsten Ansätze werden anhand von Google Trends Daten illustriert. Darüber hinaus werfen wir einen Blick auf aktuelle Forschungsströme und zeigen noch offene Forschungsfragen auf.

Time series is a common and well-known way for describing temporal data. However, most of the state-of-the-art techniques for analysing time series have focused on generating a representation for a single level of resolution. For analysing of a time series at several levels of resolutions, one would require to compute different representations, one for each resolution level. We introduce a multi-resolution representation for time series based on local trends and mean values. We require the level of resolution as parameter, but it can be automatically computed if we consider the maximum resolution of the time series. Our technique represents a time series using trend-value pairs on each segment belonging to a resolution level. To provide a useful representation for data mining tasks, we also propose dissimilarity measures and a symbolic representation based on the SAX technique for efficient similarity search using a multi-resolution indexing scheme. We evaluate our method for classification and discord discovery tasks over a diversity of data domains, achieving a better performance in terms of efficiency and effectiveness compared with some of the best-known classic techniques. Indeed, for some of the experiments, the time series mining algorithms using our multi-resolution representation were an order of magnitude faster, in terms of distance computations, than the state of the art.

Time series data are ubiquitous and being generated at an unprecedented speed and volume in many fields including finance, medicine, oil and gas industry and other business domains. Many techniques have been developed to analyze time series and understand the system that produces them. In this paper we propose a hybrid approach to improve the accuracy of time series classifiers by using Hidden Markov Models (HMM). The proposed approach is based on the principle of learning by mistakes. A HMM model is trained using the confusion matrices which are normally used to measure the classification accuracy. Misclassified samples are the basis of learning process. Our approach improves the classification accuracy by executing a second cycle of classification taking into account the temporal relations in the data. The objective of the proposed approach is to utilize the strengths of Hidden Markov Models (dealing with temporal data) to complement the weaknesses of other classification techniques. Consequently, instead of finding single isolated patterns, we focus on understanding the relationships between these patterns. The proposed approach was evaluated with a case study. The target of the case study was to classify real drilling data generated by rig sensors. Experimental evaluation proves the feasibility and effectiveness of the approach.

This work presents a novel approach to multivariate time series classification. The method exploits the multivariate structure of the time series and the possibilities of the stacking ensemble method. The basics of the method may be described in three steps: first, decomposing the multivariate time series on its constituent univariate time series; second, inducing a classifier for each univariate time series plus and additional multivariate classifier for the whole time series; third, creating the final multivariate time series classifier stacking the previous classifiers. The ensemble obtained has the potential to improve the accuracy of the single multivariate time series classifier. Several configurations of the stacking method have been tested on seven multivariate time series data sets. In five out of seven data sets, the proposed method obtains the smallest error rate. Moreover, in two out of seven data sets, stacking only the univariate time series classifiers provides the best results. The experimental results show that when a multivariate time series method does not produce an accurate classifier, stacking it with univariate time series classifiers is an alternative worthy of consideration.

This paper takes a fundamental new approach to symbolic time series analysis of real time data acquired from human driven mining equipment, which can be seen as stochastic physical systems with non analytic human interaction (hybrid systems). The developed framework uses linear differential operators (LDO) to include the system dynamics within the analysis, whereas the metaphor of language is used to mimic the human interaction. After applying LDO, the multidimensional data stream is converted into a single symbolic time series yielding a more abstract but highly condense representation of the original data. Inspired by natural language, the presented algorithm combines iteratively symbol pairs (word pairs) which occur frequently to new symbols/words; a machine-specific language emerges in a hierarchical manner, which automatically structures the dataset into segments and sub-segments. As a demonstration, the automatic recognition of operation modes of a bucket-wheel excavator is presented, proving the metaphor of language to be valuable in such hybrid systems.

This paper presents a novel method to detect anomaly using Fuzzy AdaBoost and Dynamic Naive Bayesian classifier. Dynamic Naive Bayesian Classifier (DNBC) is an extension of Hidden Markov Models (HMM) that is used here to model multivariate observation sequences typically generated from multiple sensors associated to monitoring processes. The Fuzzy AdaBoost (FAB) method is used for ensembling multiple DNBCs to classify an instance as an anomaly. FAB needs Footprint of Uncertainty (FOU) of error that is further used to Figure out the misclassification error and update weights of the data samples required during the boosting process. Here, we introduce an approach to initialize the intervals of FOU using the statistical assets of data that belongs to the normal class. The efficacy of the proposed method is demonstrated through a case study on a stuck pipe problem that occurs during oil well drilling process.

This paper presents a hybrid classifier that combines a Long Short Term Memory (LSTM) and an Evidence Feed Forward Hidden Markov Model (EFF-HMM) to classify multivariate time series (MTS). Learning of the EFF-HMM is performed based on mistakes of the LSTM. Confusion matrix obtained after classification of the MTS by the LSTM is employed during a learning process of the EFF-HMM. The EFF-HMM efficiently models a temporal characteristic and uncertainty of the MTS. The hybrid classifier combines strengths of the LSTM and EFFHMM to enhance the accuracy of the MTS classification. The proposed method is tested on Human Activity Recognition (HAR) dataset to classify various human activities. The experiments and results show the proposed method outperforms as compared to state of the art methods.

This paper discusses the Quality of Service metrics in the integration of FHAMIPv6/MPLS/Diffserv protocols. These metrics are analyzed using the load balancing algorithm in the presence of network congestion. The metrics used are delay, jitter, and throughput. These metrics are analyzed when a handoff occurs in the ad hoc network. The results obtained show that the load balancing algorithm is a great option in the integration to optimize the Quality of Service in ad hoc and hybrid network when a handoff occurs.

Theoretical aspects of data mining.- Applications of data mining.-Multimedia data.- Marketing.- Medicine.- Agriculture.- Process control.- Industry and society.

Theoretical aspects of data mining.- Applications of data mining.-Multimedia data.- Marketing.- Medicine.- Agriculture.- Process control.- Industry and society.

The scope of this paper is to review and evaluate all constant round Group Key Agreement (GKA) protocols proposed so far in the literature. We have gathered all GKA protocols that require 1,2,3,4 and 5 rounds and examined their efficiency. In particular, we calculated each protocol's computation and communication complexity and using proper assessments we compared their total energy cost. The evaluation of all protocols, interesting on its own, can also serve as a reference point for future works and contribute to the establishment of new, more efficient constant round protocols.

The paper presents new machine learning methods: signal composition, which classifies time-series regardless of length, type, and quantity; and self-labeling, a supervised-learning enhancement. The paper describes further the implementation of the methods on a financial search engine system to identify behavioral similarities among time-series representing monthly returns of 11,312 hedge funds operated during approximately one decade (2000 - 2010). The presented approach of cross-category and cross-location classification assists the investor to identify alternative investments.

The paper presents a novel agile approach to large scale design of enterprise data warehouses based on a Data Vault model. An original, simple and direct algorithm is defined for the incremental design of physical Data Vault type enterprise data warehouses, using source data meta-model and rules, and used in developing a prototype case tool for Data Vault design. This approach solves primary requirements for a system of record, that is, preservation of all source information, and fully addresses flexibility and scalability expectations. Our approach benefits from Data Vault dependencies minimizations and rapid loads opportunities enabling greatly simplified ETL transformations in a way not possible with traditional (i.e. non data vault based) data warehouse designs. The approach is illustrated using a realistic example from the healthcare domain.

Subspace clustering detects the clusters that are existing in the subspaces of the feature space. Density based subspace clustering defines clusters as regions of high density existing in subspaces of multidimensional datasets. This paper discusses the concept of closed interesting subspaces under density divergence context for multivariate datasets and proposes an algorithm to transform the multivariate time series to a symbol sequence using the closed interesting subspaces. The proposed transformation allows the applicability of any of the symbolic sequential mining algorithms to efficiently extract sequential patterns which capture the interdependencies and co-variations among groups of time series variables. The multivariate time series transformation technique is explained using a sample dataset. It is evaluated using a real world weather dataset obtained from Cambridge University. The representation power of the closed interesting subspaces and maximal interesting subspaces in transforming multivariate time series is compared.

Sub-ontology, as the name implies, is a subset of an ontology tailored for a specific domain. Large-scale ontology data can be optimized by extracting the required sub-ontology. Sub-ontology extraction that reflects the user requirements is one method of addressing the issue. Here, sub-ontology extraction methods are called sub-ontology optimization schemes. In this paper, we develop a visual view for users to adopt sub-ontology extraction methods with a particular aim of providing users a front-end tool to perform a particular sub-ontology extraction. Likewise, we construct a Semantic Grid prototype environment to extract the sub-ontology and conducted preliminary evaluation through illustrations using the UMLS Semantic Network. Functional evaluations provided promising results to pursue further experiments and studies on the problem issue.

Segmentation and labeling of different activities in multivariate time series data is an important task in many domains. There is a multitude of automatic segmentation and labeling methods available, which are designed to handle different situations. These methods can be used with multiple parametrizations, which leads to an overwhelming amount of options to choose from. To this end, we present a conceptual design of a Visual Analytics framework (1) to select appropriate segmentation and labeling methods with appropriate parametrizations, (2) to analyze the (multiple) results, (3) to understand different kinds and origins of uncertainties in these results, and (4) to reason which methods and which parametrizations yield stable results and fine-tune these configurations if necessary.

SAX Symbolic Aggregate approXimation is one of the main symbolization techniques for time series. A well-known limitation of SAX is that trends are not taken into account in the symbolization. This paper proposes 1d-SAX a method to represent a time series as a sequence of symbols that each contain information about the average and the trend of the series on a segment. We compare the efficiency of SAX and 1d-SAX in terms of goodness-of-fit, retrieval and classification performance for querying a time series database with an asymmetric scheme. The results show that 1d-SAX improves performance using equal quantity of information, especially when the compression rate increases.

SAX (Symbolic Aggregate approXimation) est une des techniques majeures de symbolisation des series temporelles. La non prise en compte des tendances dans la symbolisation est une limitation bien connue de SAX. Cet article presente 1d-SAX, une methode pour representer une serie temporelle par une sequence de symboles contenant des informations sur la moyenne et la tendance des fenetres successives de la serie segmentee. Nous comparons l'efficacite de 1d-SAX vs SAX dans une tâche de classification de series temporelles d'images satellites. Les resultats montrent que 1d-SAX ameliore les taux de classification pour une quantite d'information identique utilisee.

Research in time series (TS) knowledge discovery in general, and in similarity search in particular has been very active in recent years, due to the number of application domains that are progressively requiring to work with large amounts of high dimensional TS data. Unfortunately, for many practical applications, the high dimensionality in such frameworks makes it difficult to uncover important patterns from the raw data. Hence TS transformation techniques have become important preprocessing tools for many pattern recognition tasks. In this paper we investigate the problem of similarity search in TS and proposea symbolic transformation process that incorporates the TS value and trend information to enhance accuracy in the search results, and a symbolic similarity measure. We conduct numerous experiments to evaluate the performance of the proposed technique. Our results indicate a better capture of the time series characteristics, while providing increased accuracy and efficiency.

Recognizing patterns in time series has become a necessary machine learning task in many fields including medicine, finance, business and oil and gas industry. In this paper we propose a feature-based approach to recognize patterns in drilling time series data. Our approach consists of four phases which are: data preparation, feature extraction, feature selection and classifier training. In the first phase, the sensor-generated data required for building the recognition models are collected and prepared. In the second phase, the prepared time series data are transformed into a compact representation. The compact representation of the data consists of a set of statistical features extracted by sliding a window across the time series. In the third phase, numerous feature selection algorithms are applied to select a subset of most informative features from the statistical features set. Finally, the selected features are exploited to train a classifier that is used for final pattern recognition. Numerous feature weighting and selection algorithms were tested to find which statistical measures clearly distinguish between several different patterns. In addition, many classification techniques were employed to find the best one in terms of accuracy and speed. Experimental evaluation with real data shows that our approach has the ability to extract and select the best features and build accurate classifiers. Four different real-world drilling scenarios were used in the experiments. The performance of the classifiers was evaluated by using the cross-validation method.