Modelling semantics across multiple time series and its applications

Analysis based on the holistic multiple time series system has been a practical and crucial topic. In this paper, we mainly study a new problem that how the data is produced underneath the multiple time series system, which means how to model time series data generating and evolving rules (here denoted as semantics). We assume that there exist a set of latent states, which are the system basis and make the system run: data generating and evolving. Thus, there are several challenges on the problem: (1) How to detect the latent states; (2) How to learn the rules based on the states; (3) What the semantics can be used for. Hence, a novel correlation field-based semantics learning method is proposed to learn the semantics. In the method, we first detect latent state assignment by comprehensively considering kinds of multiple time series characteristics, which contain tick-by-tick data, temporal ordering, relationship among multiple time series and so on. Then, the semantics are learnt by Bayesian Markov characteristic. Actually, the learned semantics could be applied into various applications, such as prediction or anomaly detection for further analysis. Thus, we propose two algorithms based on the semantics knowledge, which are applied to make next-n step prediction and detect anomalies respectively. Some experiments on real world data sets were conducted to show the efficiency of our proposed method.

DDR: an index method for large time-series datasets

The tree index structure is a traditional method for searching similar data in large datasets. It is based on the presupposition that most sub-trees are pruned in the searching process. As a result, the number of page accesses is reduced. However, time-series datasets generally have a very high dimensionality. Because of the so-called dimensionality curse, the pruning effectiveness is reduced in high dimensionality. Consequently, the tree index structure is not a suitable method for time-series datasets. In this paper, we propose a two-phase (filtering and refinement) method for searching time-series datasets. In the filtering step, a quantizing time-series is used to construct a compact file which is scanned for filtering out irrelevant. A small set of candidates is translated to the second step for refinement. In this step, we introduce an effective index compression method named grid-based datawise dimensionality reduction (DRR) which attempts to preserve the characteristics of the time-series. An experimental comparison with existing techniques demonstrates the utility of our approach.

Labour productivity trends in Australian manufacturing : some time series properties

Labour productivity plays a significant role in economic growth, labour demand and employment situation of a particular economy. In this light, the presence of a structural break in productivity, and its unit root property, has important consequences for the overall economy and in major sectors such as manufacturing. In this article, using some recently developed unit root tests, we examine: (i) the null hypothesis of a unit root in the log-level of labour productivity for 38 manufacturing subdivisions against the alternative of trend stationarity over a three-decade period; and (ii) the presence of a structural break in the series, and whether the break has had a permanent or a transitory effect on manufacturing labour productivity. Our main finding is that shocks to labour productivity have had a transitory effect, implying that policies are likely to have only short-term effects.

Structural breaks and unit roots in Australian macroeconomic time series

We examine the unit root properties of 16 Australian macroeconomic time series using monthly data spanning the period 1960–2004. In addition to the standard Augmented Dickey Fuller (ADF) test, we implement one- and two-break endogenous structural break ADF-type unit root tests as well as one- and two-break Lagrange multiplier (LM) unit root tests. While the ADF test provides relatively little evidence against the unit root null hypothesis, once we allow for structural breaks we are able to reject the unit root null for just under half of the variables at the 10% level or better.

What determines bilateral aid allocations? New evidence for time series data

This paper models the allocation of bilateral foreign development aid to developing countries. A simple theoretical framework is developed, in which aid is treated as a private good of a donor country bureaucratic group responsible for bilateral aid allocation. This model is applied to time series data for ten principal recipients of bilateral official development assistance. Features of this application are that it caters for the joint determination of aid allocations and for donor allocation behavior to differ among individual recipient countries. Results indicate that both recipient need and donor interest variables determine the amount of foreign aid to developing countries, and that donor allocation behavior often differs markedly among recipients.

Integrating time-series hydroacoustics and video observations for detecting changes in benthic habitats

The ability to quantify change in marine benthic habitats must be considered a key goal of marine habitat mapping activities. Changes in distribution of distinct suites of benthic biological species may occur as a result of natural or human induced processes and these processes may operate at a range of temporal and spatial scales. It is important to understand natural small scale inter-annual patterns of change in order to separate these signals from potential patterns of longer term change. Work to describe these processes of change from an acoustic remote sensing stand point has thus far been limited due to the relatively recent availability of full coverage swath acoustic datasets and cost pressures associated with multiple surveys of the same area. This paper describes the use of landscape transition analysis as a means to differentiate seemingly random patterns of habitat change from systematic signals of habitat transition at a shallow (10–50 m depth) 18 km2 study area on the temperate Australian continental shelf between the years 2006 and 2007. Supervised classifications for each year were accomplished using independently collected high resolution (3 m cell-size) multibeam echosounder (MBES) and video-derived reference data. Of the 4 representative biotic classes considered, signals of directional systematic changes were observed to occur between a shallow kelp dominated class, a deep sessile invertebrate dominated class and a mixed class of kelp and sessile invertebrates. These signals of change are interpreted as inter-annual variation in the density and depth related extent of canopy forming kelp species at the site, a phenomenon reported in smaller scale temporal studies of the same species. The methods applied in this study provide a detailed analysis of the various components of the traditional change detection cross tabulation matrix allowing identification of the strongest signals of systematic habitat transitions across broad geographical regions. Identifying clear patterns of habitat change is an important first step in linking these patterns to the processes that drive them.

The application of an ensemble of boosted elman networks to time series prediction : a benchmark study

In this paper, the application of multiple Elman neural networks to time series data regression problems is studied. An ensemble of Elman networks is formed by boosting to enhance the performance of the individual networks. A modified version of the AdaBoost algorithm is employed to integrate the predictions from multiple networks. Two benchmark time series data sets, i.e., the Sunspot and Box-Jenkins gas furnace problems, are used to assess the effectiveness of the proposed system. The simulation results reveal that an ensemble of boosted Elman networks can achieve a higher degree of generalization as well as performance than that of the individual networks. The results are compared with those from other learning systems, and implications of the performance are discussed.

Faster and parameter-free discord search in quasi-periodic time series

Time series discord has proven to be a useful concept for time-series anomaly identification. To search for discords, various algorithms have been developed. Most of these algorithms rely on pre-building an index (such as a trie) for subsequences. Users of these algorithms are typically required to choose optimal values for word-length and/or alphabet-size parameters of the index, which are not intuitive. In this paper, we propose an algorithm to directly search for the top-K discords, without the requirement of building an index or tuning external parameters. The algorithm exploits quasi-periodicity present in many time series. For quasi-periodic time series, the algorithm gains significant speedup by reducing the number of calls to the distance function.