Multiple time series anomaly detection based on compression and correlation analysis: a medical surveillance case study

In this paper, we present a novel anomaly detection framework for multiple heterogeneous yet correlated time series, such as the medical surveillance series data. In our framework, we propose an anomaly detection algorithm from the viewpoint of trend and correlation analysis. Moreover, to efficiently process huge amount of observed time series, a new clustering-based compression method is proposed. Experimental results indicate that our framework is more effective and efficient than its peers. © 2012 Springer-Verlag Berlin Heidelberg.

Parallel Time Series Analysis and Income Inequality: A New Approach to Comparative Case Studies

This case study deals with the role of time series analysis in sociology, and its relationship with the wider literature and methodology of comparative case study research. Time series analysis is now well-represented in top-ranked sociology journals, often in the form of ‘pooled time series’ research designs. These studies typically pool multiple countries together into a pooled time series cross-section panel, in order to provide a larger sample for more robust and comprehensive analysis. This approach is well suited to exploring trans-national phenomena, and for elaborating useful macro-level theories specific to social structures, national policies, and long-term historical processes. It is less suited however, to understanding how these global social processes work in different countries. As such, the complexities of individual countries - which often display very different or contradictory dynamics than those suggested in pooled studies – are subsumed. Meanwhile, a robust literature on comparative case-based methods exists in the social sciences, where researchers focus on differences between cases, and the complex ways in which they co-evolve or diverge over time. A good example of this is the inequality literature, where although panel studies suggest a general trend of rising inequality driven by the weakening power of labour, marketisation of welfare, and the rising power of capital, some countries have still managed to remain resilient. This case study takes a closer look at what can be learned by applying the insights of case-based comparative research to the method of time series analysis. Taking international income inequality as its point of departure, it argues that we have much to learn about the viability of different combinations of policy options by examining how they work in different countries over time. By taking representative cases from different welfare systems (liberal, social democratic, corporatist, or antipodean), we can better sharpen our theories of how policies can be more specifically engineered to offset rising inequality. This involves a fundamental realignment of the strategy of time series analysis, grounding it instead in a qualitative appreciation of the historical context of cases, as a basis for comparing effects between different countries.

Financial time series analysis : Chaos and neurodynamics approach

This work aims at combining the Chaos theory postulates and Artificial Neural Networks classification and predictive capability, in the field of financial time series prediction. Chaos theory, provides valuable qualitative and quantitative tools to decide on the predictability of a chaotic system. Quantitative measurements based on Chaos theory, are used, to decide a-priori whether a time series, or a portion of a time series is predictable, while Chaos theory based qualitative tools are used to provide further observations and analysis on the predictability, in cases where measurements provide negative answers. Phase space reconstruction is achieved by time delay embedding resulting in multiple embedded vectors. The cognitive approach suggested, is inspired by the capability of some chartists to predict the direction of an index by looking at the price time series. Thus, in this work, the calculation of the embedding dimension and the separation, in Takens‘ embedding theorem for phase space reconstruction, is not limited to False Nearest Neighbor, Differential Entropy or other specific method, rather, this work is interested in all embedding dimensions and separations that are regarded as different ways of looking at a time series by different chartists, based on their expectations. Prior to the prediction, the embedded vectors of the phase space are classified with Fuzzy-ART, then, for each class a back propagation Neural Network is trained to predict the last element of each vector, whereas all previous elements of a vector are used as features.

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.

Development of record-breaking statistics for climatological time-series analysis

The number of record-breaking events expected to occur in a strictly stationary time-series depends only on the number of values in the time-series, regardless of distribution. This holds whether the events are record-breaking highs or lows and whether we count from past to present or present to past. However, these symmetries are broken in distinct ways by trends in the mean and variance. We define indices that capture this information and use them to detect weak trends from multiple time-series. Here, we use these methods to answer the following questions: (1) Is there a variability trend among globally distributed surface temperature time-series? We find a significant decreasing variability over the past century for the Global Historical Climatology Network (GHCN). This corresponds to about a 10% change in the standard deviation of inter-annual monthly mean temperature distributions. (2) How are record-breaking high and low surface temperatures in the United States affected by time period? We investigate the United States Historical Climatology Network (USHCN) and find that the ratio of record-breaking highs to lows in 2006 increases as the time-series extend further into the past. When we consider the ratio as it evolves with respect to a fixed start year, we find it is strongly correlated with the ensemble mean. We also compare the ratios for USHCN and GHCN (minus USHCN stations). We find the ratios grow monotonically in the GHCN data set, but not in the USHCN data set. (3) Do we detect either mean or variance trends in annual precipitation within the United States? We find that the total annual and monthly precipitation in the United States (USHCN) has increased over the past century. Evidence for a trend in variance is inconclusive.

TIME SERIES ANALYSIS AS INPUT FOR PREDICTIVE MODELING: PREDICTING CARDIAC ARREST IN A PEDIATRIC INTENSIVE CARE UNIT

The first manuscript, entitled "Time-Series Analysis as Input for Clinical Predictive Modeling: Modeling Cardiac Arrest in a Pediatric ICU" lays out the theoretical background for the project. There are several core concepts presented in this paper. First, traditional multivariate models (where each variable is represented by only one value) provide single point-in-time snapshots of patient status: they are incapable of characterizing deterioration. Since deterioration is consistently identified as a precursor to cardiac arrests, we maintain that the traditional multivariate paradigm is insufficient for predicting arrests. We identify time series analysis as a method capable of characterizing deterioration in an objective, mathematical fashion, and describe how to build a general foundation for predictive modeling using time series analysis results as latent variables. Building a solid foundation for any given modeling task involves addressing a number of issues during the design phase. These include selecting the proper candidate features on which to base the model, and selecting the most appropriate tool to measure them. We also identified several unique design issues that are introduced when time series data elements are added to the set of candidate features. One such issue is in defining the duration and resolution of time series elements required to sufficiently characterize the time series phenomena being considered as candidate features for the predictive model. Once the duration and resolution are established, there must also be explicit mathematical or statistical operations that produce the time series analysis result to be used as a latent candidate feature. In synthesizing the comprehensive framework for building a predictive model based on time series data elements, we identified at least four classes of data that can be used in the model design. The first two classes are shared with traditional multivariate models: multivariate data and clinical latent features. Multivariate data is represented by the standard one value per variable paradigm and is widely employed in a host of clinical models and tools. These are often represented by a number present in a given cell of a table. Clinical latent features derived, rather than directly measured, data elements that more accurately represent a particular clinical phenomenon than any of the directly measured data elements in isolation. The second two classes are unique to the time series data elements. The first of these is the raw data elements. These are represented by multiple values per variable, and constitute the measured observations that are typically available to end users when they review time series data. These are often represented as dots on a graph. The final class of data results from performing time series analysis. This class of data represents the fundamental concept on which our hypothesis is based. The specific statistical or mathematical operations are up to the modeler to determine, but we generally recommend that a variety of analyses be performed in order to maximize the likelihood that a representation of the time series data elements is produced that is able to distinguish between two or more classes of outcomes. The second manuscript, entitled "Building Clinical Prediction Models Using Time Series Data: Modeling Cardiac Arrest in a Pediatric ICU" provides a detailed description, start to finish, of the methods required to prepare the data, build, and validate a predictive model that uses the time series data elements determined in the first paper. One of the fundamental tenets of the second paper is that manual implementations of time series based models are unfeasible due to the relatively large number of data elements and the complexity of preprocessing that must occur before data can be presented to the model. Each of the seventeen steps is analyzed from the perspective of how it may be automated, when necessary. We identify the general objectives and available strategies of each of the steps, and we present our rationale for choosing a specific strategy for each step in the case of predicting cardiac arrest in a pediatric intensive care unit. Another issue brought to light by the second paper is that the individual steps required to use time series data for predictive modeling are more numerous and more complex than those used for modeling with traditional multivariate data. Even after complexities attributable to the design phase (addressed in our first paper) have been accounted for, the management and manipulation of the time series elements (the preprocessing steps in particular) are issues that are not present in a traditional multivariate modeling paradigm. In our methods, we present the issues that arise from the time series data elements: defining a reference time; imputing and reducing time series data in order to conform to a predefined structure that was specified during the design phase; and normalizing variable families rather than individual variable instances. The final manuscript, entitled: "Using Time-Series Analysis to Predict Cardiac Arrest in a Pediatric Intensive Care Unit" presents the results that were obtained by applying the theoretical construct and its associated methods (detailed in the first two papers) to the case of cardiac arrest prediction in a pediatric intensive care unit. Our results showed that utilizing the trend analysis from the time series data elements reduced the number of classification errors by 73%. The area under the Receiver Operating Characteristic curve increased from a baseline of 87% to 98% by including the trend analysis. In addition to the performance measures, we were also able to demonstrate that adding raw time series data elements without their associated trend analyses improved classification accuracy as compared to the baseline multivariate model, but diminished classification accuracy as compared to when just the trend analysis features were added (ie, without adding the raw time series data elements). We believe this phenomenon was largely attributable to overfitting, which is known to increase as the ratio of candidate features to class examples rises. Furthermore, although we employed several feature reduction strategies to counteract the overfitting problem, they failed to improve the performance beyond that which was achieved by exclusion of the raw time series elements. Finally, our data demonstrated that pulse oximetry and systolic blood pressure readings tend to start diminishing about 10-20 minutes before an arrest, whereas heart rates tend to diminish rapidly less than 5 minutes before an arrest.

Time series analysis of a Web search engine transaction log

In this paper, we use time series analysis to evaluate predictive scenarios using search engine transactional logs. Our goal is to develop models for the analysis of searchers’ behaviors over time and investigate if time series analysis is a valid method for predicting relationships between searcher actions. Time series analysis is a method often used to understand the underlying characteristics of temporal data in order to make forecasts. In this study, we used a Web search engine transactional log and time series analysis to investigate users’ actions. We conducted our analysis in two phases. In the initial phase, we employed a basic analysis and found that 10% of searchers clicked on sponsored links. However, from 22:00 to 24:00, searchers almost exclusively clicked on the organic links, with almost no clicks on sponsored links. In the second and more extensive phase, we used a one-step prediction time series analysis method along with a transfer function method. The period rarely affects navigational and transactional queries, while rates for transactional queries vary during different periods. Our results show that the average length of a searcher session is approximately 2.9 interactions and that this average is consistent across time periods. Most importantly, our findings shows that searchers who submit the shortest queries (i.e., in number of terms) click on highest ranked results. We discuss implications, including predictive value, and future research.

One-year delayed effect of fog on malaria transmission : a time-series analysis in the rain forest area of Mengla County, south-west China

Background: Malaria is a major public health burden in the tropics with the potential to significantly increase in response to climate change. Analyses of data from the recent past can elucidate how short-term variations in weather factors affect malaria transmission. This study explored the impact of climate variability on the transmission of malaria in the tropical rain forest area of Mengla County, south-west China. Methods: Ecological time-series analysis was performed on data collected between 1971 and 1999. Auto-regressive integrated moving average (ARIMA) models were used to evaluate the relationship between weather factors and malaria incidence. Results: At the time scale of months, the predictors for malaria incidence included: minimum temperature, maximum temperature, and fog day frequency. The effect of minimum temperature on malaria incidence was greater in the cool months than in the hot months. The fog day frequency in October had a positive effect on malaria incidence in May of the following year. At the time scale of years, the annual fog day frequency was the only weather predictor of the annual incidence of malaria. Conclusion: Fog day frequency was for the first time found to be a predictor of malaria incidence in a rain forest area. The one-year delayed effect of fog on malaria transmission may involve providing water input and maintaining aquatic breeding sites for mosquitoes in vulnerable times when there is little rainfall in the 6-month dry seasons. These findings should be considered in the prediction of future patterns of malaria for similar tropical rain forest areas worldwide.

Nitrous oxide emissions from subtropical horticultural soils : a time series analysis

Time series regression models were used to examine the influence of environmental factors (soil water content and soil temperature) on the emissions of nitrous oxide (N2O) from subtropical soils, by taking into account temporal lagged environmental factors, autoregressive processes, and seasonality for three horticultural crops in a subtropical region of Australia. Fluxes of N2O, soil water content, and soil temperature were determined simultaneously on a weekly basis over a 12-month period in South East Queensland. Annual N2O emissions for soils under mango, pineapple, and custard apple were 1590, 1156, and 2038 g N2O-N/ha, respectively, with most emissions attributed to nitrification. The N2O-N emitted from the pineapple and custard apple crops was equivalent to 0.26 and 2.22%, respectively, of the applied mineral N. The change in soil water content was the key variable for describing N2O emissions at the weekly time-scale, with soil temperature at a lag of 1 month having a significant influence on average N2O emissions (averaged) at the monthly time-scale across the three crops. After accounting for soil temperature and soil water content, both the weekly and monthly time series regression models exhibited significant autocorrelation at lags of 1–2 weeks and 1–2 months, and significant seasonality for weekly N2O emissions for mango crop and for monthly N2O emissions for mango and custard apple crops in this location over this time-frame. Time series regression models can explain a higher percentage of the temporal variation of N2O emission compared with simple regression models using soil temperature and soil water content as drivers. Taking into account seasonal variability and temporal persistence in N2O emissions associated with soil water content and soil temperature may lead to a reduction in the uncertainty surrounding estimates of N2O emissions based on limited sampling effort.

Socio-demographic vulnerability to heatwave impacts in Brisbane, Australia : a time series analysis

Objective: Examining the association between socioeconomic disadvantage and heat-related emergency department (ED) visits during heatwave periods in Brisbane, 2000–2008. Methods: Data from 10 public EDs were analysed using a generalised additive model for disease categories, age groups and gender. Results: Cumulative relative risks (RR) for non-external causes other than cardiovascular and respiratory diseases were 1.11 and 1.05 in most and least disadvantaged areas, respectively. The pattern persisted on lags 0–2. Elevated risks were observed for all age groups above 15 years in all areas. However, with RRs of 1.19–1.28, the 65–74 years age group in more disadvantaged areas stood out, compared with RR=1.08 in less disadvantaged areas. This pattern was observed on lag 0 but did not persist. The RRs for male presentations were 1.10 and 1.04 in most and less disadvantaged areas; for females, RR was 1.04 in less disadvantaged areas. This pattern persisted across lags 0–2. Conclusions: Heat-related ED visits increased during heatwaves. However, due to overlapping confidence intervals, variations across socioeconomic areas should be interpreted cautiously. Implications: ED data may be utilised for monitoring heat-related health impacts, particularly on the first day of heatwaves, to facilitate prompt interventions and targeted resource allocation.

A time series analysis of presentations to Queensland health facilities for alcohol-related conditions, following the increase in 'alcopops' tax

Objective: In response to concerns about the health consequences of high-risk drinking by young people, the Australian Government increased the tax on pre-mixed alcoholic beverages ('alcopops') favoured by this demographic. We measured changes in admissions for alcohol-related harm to health throughout Queensland, before and after the tax increase in April 2008. Methods: We used data from the Queensland Trauma Register, Hospitals Admitted Patients Data Collection, and the Emergency Department Information System to calculate alcohol-related admission rates per 100,000 people, for 15 - 29 year-olds. We analysed data over 3 years (April 2006 - April 2009), using interrupted time-series analyses. This covered 2 years before, and 1 year after, the tax increase. We investigated both mental and behavioural consequences (via F10 codes), and intentional/unintentional injuries (S and T codes). Results: We fitted an auto-regressive integrated moving average (ARIMA) model, to test for any changes following the increased tax. There was no decrease in alcohol-related admissions in 15 - 29 year-olds. We found similar results for males and females, as well as definitions of alcohol-related harms that were narrow (F10 codes only) and broad (F10, S and T codes). Conclusions: The increased tax on 'alcopops' was not associated with any reduction in hospital admissions for alcohol-related harms in Queensland 15 - 29 year-olds.

Anticipating Chinese Tourists Arrivals in Australia: A Time Series Analysis

Given the growing importance of the Chinese tourist market to Australia, an understanding of Chinese tourists' arrival patterns is essential to accurate forecasting of future arrivals. Drawing on 25 years of records (1991-2015), this study developed a time-series model of monthly arrivals of Chinese tourists in Australia. The model reflects the exponentially increasing trend and strong seasonality of arrivals. Excellent results from validation of the model's forecasts endorsed this time-series model's potential in the policy prescription and management practice of Australian tourism industries.

Stochastic Modeling of an Aircraft Traversing a Runway Using Time Series Analysis

Time series, from a narrow point of view, is a sequence of observations on a stochastic process made at discrete and equally spaced time intervals. Its future behavior can be predicted by identifying, fitting, and confirming a mathematical model. In this paper, time series analysis is applied to problems concerning runwayinduced vibrations of an aircraft. A simple mathematical model based on this technique is fitted to obtain the impulse response coefficients of an aircraft system considered as a whole for a particular type of operation. Using this model, the output which is the aircraft response can be obtained with lesser computation time for any runway profile as the input.