A time series forest for classification and feature extraction

Deng, Houtao; Runger, George C.; Tuv, Eugene; Martyanov, Vladimir

doi:10.1016/j.ins.2013.02.030

Cited by 488 publications

(281 citation statements)

References 22 publications

Supporting

Mentioning

254

Contrasting

Unclassified

Order By: Relevance

“…However, in [11] it was shown that there is little difference between the most cited measures in the literature. The majority of recent work proposing alternative classifiers [23,10,7] tends to embed any transforms within the classification process, thus making it hard to qualitatively or quantitatively assess which element of the proposed technique is the source of any improvement in accuracy. We separate the process of representation and the classifier in the hope of more transparency.…”

Section: Data Transformationsmentioning

confidence: 99%

“…Ensembles have been used in time series data mining [7,23,10]. Deng and Runger [10] propose a random forest [5] built on summary statistics (mean, slope and variance) of subseries.…”

Section: 22mentioning

confidence: 99%

“…Despite the evidence in favour of 1-NN classifiers with Euclidean or DTW distance, there has been a flurry of recent research publications proposing alternative approaches. These include shapelets [19,28], weighted dynamic time warping [15], support vector machines built on variable intervals [23], tree based ensembles constructed on summary statistics [10], and fusion of alternative distance measures [7]. These approaches combine deriving features and/or distance measures with classifiers of varying degrees of complexity, and the majority compare their algorithms to 1-NN with Euclidean or DTW distance.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Transformation Based Ensembles for Time Series Classification

Bagnall¹,

Davis²,

Hills³

et al. 2012

Proceedings of the 2012 SIAM International Conference on Data Mining

View full text Add to dashboard Cite

Until recently, the vast majority of data mining time series classification (TSC) research has focused on alternative distance measures for 1-Nearest Neighbour (1-NN) classifiers based on either the raw data, or on compressions or smoothing of the raw data. Despite the extensive evidence in favour of 1-NN classifiers with Euclidean or Dynamic Time Warping distance, there has also been a flurry of recent research publications proposing classification algorithms for TSC. Generally, these classifiers describe different ways of incorporating summary measures in the time domain into more complex classifiers. Our hypothesis is that the easiest way to gain improvement on TSC problems is simply to transform into an alternative data space where the discriminatory features are more easily detected. To test our hypothesis, we perform a range of benchmarking experiments in the time domain, before evaluating nearest neighbour classifiers on data transformed into the power spectrum, the autocorrelation function, and the principal component space. We demonstrate that on some problems there is dramatic improvement in the accuracy of classifiers built on the transformed data over classifiers built in the time domain, but that there is also a wide variance in accuracy for a particular classifier built on different data transforms. To overcome this variability, we propose a simple transformation based ensemble, then demonstrate that it improves performance and reduces the variability of classifiers built in the time domain only. Our advice to a practitioner with a real world TSC problem is to try transforms before developing a complex classifier; it is the easiest way to get a potentially large increase in accuracy, and may provide further insights into the underlying relationships that characterise the problem.

show abstract

Section: Data Transformationsmentioning

confidence: 99%

“…Ensembles have been used in time series data mining [7,23,10]. Deng and Runger [10] propose a random forest [5] built on summary statistics (mean, slope and variance) of subseries.…”

Section: 22mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Transformation Based Ensembles for Time Series Classification

Bagnall¹,

Davis²,

Hills³

et al. 2012

Proceedings of the 2012 SIAM International Conference on Data Mining

View full text Add to dashboard Cite

show abstract

“…Despite the evidence in favour of 1-NN classifiers with Euclidean or Dynamic Time Warping (DTW) distance, there has been a spate of recent research proposing alternative approaches. These include shapelets [13,20,19], weighted dynamic time warping [10], support vector machines built on variable intervals [16], tree based ensembles constructed on summary statistics [6], fusion of alternative distance measures [2], and transform-based ensembles [1]. We consider the shapelet approach one of the most promising of these new methods.…”

Section: Introductionmentioning

confidence: 99%

A shapelet transform for time series classification

Lines

Davis

Hills

et al. 2012

Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

282

233

View full text Add to dashboard Cite

The problem of time series classification (TSC), where we consider any real-valued ordered data a time series, presents a specific machine learning challenge as the ordering of variables is often crucial in finding the best discriminating features. One of the most promising recent approaches is to find shapelets within a data set. A shapelet is a time series subsequence that is identified as being representative of class membership. The original research in this field embedded the procedure of finding shapelets within a decision tree. We propose disconnecting the process of finding shapelets from the classification algorithm by proposing a shapelet transformation. We describe a means of extracting the k best shapelets from a data set in a single pass, and then use these shapelets to transform data by calculating the distances from a series to each shapelet. We demonstrate that transformation into this new data space can improve classification accuracy, whilst retaining the explanatory power provided by shapelets.

show abstract

“…These continuous attributes are recorded as discrete signals by means of digital systems. In the last decade particularly, many studies have focused on the discretization of already discrete signals for the discovery of knowledge [9][10][11][12][13][14][15]. There are two parameters that affect results in discretization: the number of cut points and their locations.…”

Section: Introductionmentioning

confidence: 99%

Time series adapted supervised fuzzy discretization: an application to ECG signals

Orhan

2016

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Abstract:In this study, a new method called supervised fuzzy discretization (SFD), which can be used without having expertise on data, is proposed for classifying time series datasets. Because an ECG signal has a partially stationary characteristic, its classification process is more difficult than it would be for completely stationary signals. On the other hand, because the method proposed can be used without having expertise on the data, comprehensive data like ECG signals are enough to introduce one such method. To prove the efficacy of the SFD, RR intervals selected from a common ECG database are used in the classification experiments. Some parameters, such as the coefficients of discretization, equal time slicing, learning rate, and momentum, are analyzed for the highest level of success in classification. A new mechanism called an inconsistency detector is suggested for increasing the level of success in supervised learning by adjusting the learning rate. The best results of the SFD method are compared with those of other studies in the same database, which hopefully establishes the proposed method as worth investigating in other areas because of its projected success.

show abstract

A time series forest for classification and feature extraction

Cited by 488 publications

References 22 publications

Transformation Based Ensembles for Time Series Classification

Transformation Based Ensembles for Time Series Classification

A shapelet transform for time series classification

Time series adapted supervised fuzzy discretization: an application to ECG signals

Contact Info

Product

Resources

About