29 research outputs found
A review and evaluation of elastic distance functions for time series clustering
No full textTime series clustering is the act of grouping time series data without recourse to a label. Algorithms that cluster time series can be classified into two groups: those that employ a time series specific distance measure and those that derive features from time series. Both approaches usually rely on traditional clustering algorithms such as k-means. Our focus is on partitional clustering algorithms that employ elastic distance measures, i.e. distances that perform some kind of realignment whilst measuring distance. We describe nine commonly used elastic distance measures and compare their performance with k-means and k-medoids clusterer. Our findings, based on experiments using the UCR time series archive, are surprising. We find that, generally, clustering with DTW distance is not better than using Euclidean distance and that distance measures that employ editing in conjunction with warping are significantly better than other approaches. We further observe that using k-medoids clusterer rather than k-means improves the clusterings for all nine elastic distance measures. One function, the move–split–merge (MSM) distance, is the best performing algorithm of this study, with time warp edit (TWE) distance a close second. Our conclusion is that MSM or TWE with k-medoids clusterer should be considered as a good alternative to DTW for clustering time series with elastic distance measures. We provide implementations, extensive results and guidance on reproducing results on the associated GitHub repository.</p
The great multivariate time series classification bake off: a review and experimental evaluation of recent algorithmic advances
No full textTime Series Classification (TSC) involves building predictive models for a discrete target variable from ordered, real valued, attributes. Over recent years, a new set of TSC algorithms have been developed which have made significant improvement over the previous state of the art. The main focus has been on univariate TSC, i.e. the problem where each case has a single series and a class label. In reality, it is more common to encounter multivariate TSC (MTSC) problems where the time series for a single case has multiple dimensions. Despite this, much less consideration has been given to MTSC than the univariate case. The UCR archive has provided a valuable resource for univariate TSC, and the lack of a standard set of test problems may explain why there has been less focus on MTSC. The UEA archive of 30 MTSC problems released in 2018 has made comparison of algorithms easier. We review recently proposed bespoke MTSC algorithms based on deep learning, shapelets and bag of words approaches. If an algorithm cannot naturally handle multivariate data, the simplest approach to adapt a univariate classifier to MTSC is to ensemble it over the multivariate dimensions. We compare the bespoke algorithms to these dimension independent approaches on the 26 of the 30 MTSC archive problems where the data are all of equal length. We demonstrate that four classifiers are significantly more accurate than the benchmark dynamic time warping algorithm and that one of these recently proposed classifiers, ROCKET, achieves significant improvement on the archive datasets in at least an order of magnitude less time than the other three
Unsupervised feature based algorithms for time series extrinsic regression
No full textTime Series Extrinsic Regression (TSER) involves using a set of training time series to form a predictive model of a continuous response variable that is not directly related to the regressor series. The TSER archive for comparing algorithms was released in 2022 with 19 problems. We increase the size of this archive to 63 problems and reproduce the previous comparison of baseline algorithms. We then extend the comparison to include a wider range of standard regressors and the latest versions of TSER models used in the previous study. We show that none of the previously evaluated regressors can outperform a regression adaptation of a standard classifier, rotation forest. We introduce two new TSER algorithms developed from related work in time series classification. FreshPRINCE is a pipeline estimator consisting of a transform into a wide range of summary features followed by a rotation forest regressor. DrCIF is a tree ensemble that creates features from summary statistics over random intervals. Our study demonstrates that both algorithms, along with InceptionTime, exhibit significantly better performance compared to the other 18 regressors tested. More importantly, DrCIF is the only one that significantly outperforms a standard rotation forest regressor
HIVE-COTE 2.0: a new meta ensemble for time series classification
No full textThe Hierarchical Vote Collective of Transformation-based Ensembles (HIVE-COTE) is a heterogeneous meta ensemble for time series classification. HIVE-COTE forms its ensemble from classifiers of multiple domains, including phase-independent shapelets, bag-of-words based dictionaries and phase-dependent intervals. Since it was first proposed in 2016, the algorithm has remained state of the art for accuracy on the UCR time series classification archive. Over time it has been incrementally updated, culminating in its current state, HIVE-COTE 1.0. During this time a number of algorithms have been proposed which match the accuracy of HIVE-COTE. We propose comprehensive changes to the HIVE-COTE algorithm which significantly improve its accuracy and usability, presenting this upgrade as HIVE-COTE 2.0. We introduce two novel classifiers, the Temporal Dictionary Ensemble (TDE) and Diverse Representation Canonical Interval Forest (DrCIF), which replace existing ensemble members. Additionally, we introduce the Arsenal, an ensemble of ROCKET classifiers as a new HIVE-COTE 2.0 constituent. We demonstrate that HIVE-COTE 2.0 is significantly more accurate than the current state of the art on 112 univariate UCR archive datasets and 26 multivariate UEA archive datasets
aeon: a Python toolkit for learning from time series
No full textaeon is a unified Python 3 library for all machine learning tasks involving time series. The package contains modules for time series forecasting, classification, extrinsic regression and clustering, as well as a variety of utilities, transformations and distance measures designed for time series data. aeon also has a number of experimental modules for tasks such as anomaly detection, similarity search and segmentation. aeon follows the scikit-learn API as much as possible to help new users and enable easy integration of aeon estimators with useful tools such as model selection and pipelines. It provides a broad library of time series algorithms, including efficient implementations of the very latest advances in research. Using a system of optional dependencies, aeon integrates a wide variety of packages into a single interface while keeping the core framework with minimal dependencies. The package is distributed under the 3-Clause BSD license and is available at https://github.com/aeon-toolkit/aeon
A hands-on introduction to time series classification and regression
No full textTime series classification and regression are rapidly evolving fields that find areas of application in all domains of machine learning and data science. This hands on tutorial will provide an accessible overview of the recent research in these fields, using code examples to introduce the process of implementing and evaluating an estimator. We will show how to easily reproduce published results and how to compare a new algorithm to state-of-the-art. Finally, we will work through real world examples from the field of Electroencephalogram (EEG) classification and regression. EEG machine learning tasks arise in medicine, brain-computer interface research and psychology. We use these problems to how to compare algorithms on problems from a single domain and how to deal with data with different characteristics, such as missing values, unequal length and high dimensionality. The latest advances in the fields of time series classification and regression are all available through the aeon toolkit, an open source, scikit-learn compatible framework for time series machine learning which we use to provide our code examples.</p
Detection of nuisance call centres using improved hybrid time series classification algorithms
No full textNuisance calling has become a major problem in the last couple of decades with the advent of digital VoIP and increasing automation. Millions of these nuisance calls are made every month, wasting the time of those who receive them or aiming to scam vulnerable consumers. To take action against call centres making these calls, reliably detecting that a Calling Line Identification (CLI) number is carrying nuisance traffic and identifying the source using the CLI is key. This must be done carefully, however. Phone calls and call records contain sensitive information, and consumer privacy must be taken into account and protected.
Time Series Classification (TSC) is a field of research focusing on classifying data in the format of ordered series. We further develop and investigate the use of these algorithms for detecting nuisance call centres. A high accuracy TSC algorithm, HIVE-COTE, is a hybrid heterogenous ensemble of TSC approaches, with each ensemble member extracting a different type of feature from a time series. While HIVE-COTE has shown to generally perform well on the UCR archive of time series datasets, it is slow and many algorithms have caught up to its performance as the field has developed.
In this thesis, we answer two research questions. The first question is whether we can detect nuisance traffic from call centres with minimal information using time series data and TSC algorithms. The second is whether we can improve HIVE-COTE through updating the members of the ensemble which have fallen behind in performance and scalability.
We investigate improvements to HIVE-COTE through introducing more accurate and scalable alternatives to its phase-independent interval and bag-of-words dictionary based constituent classifiers, and altering the way it generates accuracy estimates. This results in the HIVE-COTE 2.0 (HC2) algorithm. We demonstrate the HC2 is much more usable than the original HIVE-COTE for a wide variety of datasets, and once again performs as a state-of-the-art classifier in terms of accuracy on the UCR and UEA time series archives.
With our improved TSC algorithms, we classify time series data from nuisance and legitimate call centres. Given a CLI number, we format the call volume and call duration of calls made from a CLI into a time series. This approach requires no identifying information from any consumer or call centre using the CLI, while providing a profile of how calls are being made by that number. We show that our algorithms can perfectly separate legitimate and nuisance call centres using data provided by British Telecom (BT). An investigation into classifying time series with as few calls as possible shows it is still possible to achieve 100% accuracy with only a portion of the day’s traffic, allowing action to be taken in real-time
Not enough science or not enough learning? Exploring the gaps between leadership theory and practice
No full textThis paper addresses the relationships between leadership theory, practice and development, drawing on both the higher education and wider leadership literature. It explores why challenges and problems exist within the contested field of leadership theory and why gaps remain between theory and practice after more than a century of research – and indeed, with increasing levels of research, scholarship and development in the last 25 years. After highlighting the importance of context for theory, practice and development, the first section of the paper examines a range of factors that contribute to theoretical ‘contests’ including different starting assumptions made by researchers, the different focus of studies, examination of different causal links to explain leadership, differences in values and cultural lenses and different constructs, terminology and perspectives. The second section examines the challenges faced by leadership practitioners, as individuals, and through exercising leadership as a collective responsibility in the context of changing operating environments within higher education institutions and across sectors and countries. The author highlights three areas where some re-thinking of the links between theory and practice are necessary – at the input stage, linking research findings and recruitment practices; in terms of outcomes, by researching links between leaders, leadership and performance; and in process terms, to examine more deeply complex and relational dynamic of leadership in action. The third section offers a number of specific suggestions as to how closer alignment between theory, practice and development can be achieved. The paper concludes by arguing for greater maturity (in research, practice and development) that acknowledges that leadership is played out in complex, dynamic and changing social systems. A stronger emphasis on ‘leadership learning’ should deliver both better science and better outcomes for leaders and led in higher education
A detailed study of the reflection nebula, NGC 7023
No full textPolarisation and intensity maps in three broad wavebands are presented for the reflection nebula NGC7023. The data are used to investigate the structure, dust distribution and grain characteristics of the material surrounding the central illuminating star HD200775 of the reflection nebula. Calculations have been made, using a Monte-Carlo technique, for various parameters representing the structure and content of the nebula to predict and explain the observed measurements. The successful description of the observations puts severe restrictions on the nebular parameters. It is found that the geometry of the nebula is in the form of an extended cloud with a foreground conical cavity in which the illuminating star is situated. The dust grains are required to have a power law size distribution of the form n(a) = a(^4.05) and grain material corresponding to silicates is most likely although ice cannot be excluded
The Canonical Interval Forest {(CIF)} Classifier for Time Series Classification
No full textTime series classification (TSC) is home to a number of algorithm groups that utilise different kinds of discriminatory patterns. One of these groups describes classifiers that predict using phase dependant intervals. The time series forest (TSF) classifier is one of the most well known interval methods, and has demonstrated strong performance as well as relative speed in training and predictions. However, recent advances in other approaches have left TSF behind. TSF originally summarises intervals using three simple summary statistics. The `catch22' feature set of 22 time series features was recently proposed to aid time series analysis through a concise set of diverse and informative descriptive characteristics. We propose combining TSF and catch22 to form a new classifier, the Canonical Interval Forest (CIF). We outline additional enhancements to the training procedure, and extend the classifier to include multivariate classification capabilities. We demonstrate a large and significant improvement in accuracy over both TSF and catch22, and show it to be on par with top performers from other algorithmic classes. By upgrading the interval-based component from TSF to CIF, we also demonstrate a significant improvement in the hierarchical vote collective of transformation-based ensembles (HIVE-COTE) that combines different time series representations. HIVE-COTE using CIF is significantly more accurate on the UCR archive than any other classifier we are aware of and represents a new state of the art for TSC
