Change Point Detection Research Articles

Global Warming: Predictive Models and Correlation Analysis

In recent years, the Earth's rising temperatures and the shrinking Antarctic ice cover have become increasingly evident. Understanding the factors driving global warming and developing strategies to mitigate it are critical. By applying the Pettitt change-point detection method, we analyzed March temperatures from the past decade and identified a significant shift occurring in March 2015. Forecasting models, including ARIMA and LSTM, predict similar temperature trends, with the LSTM model demonstrating slightly higher accuracy. Through grey correlation analysis and multiple linear regression, we identified CO2 as the most influential factor affecting global temperatures. Therefore, reducing CO2 emissions is crucial to mitigating global warming.

Change point detection in high dimensional covariance matrix using Pillai’s statistics

AbstractThis research proposes a method to test and estimate change points in the covariance structure of high-dimensional multivariate series data. Our method uses the trace of the beta matrix, known as Pillai’s statistics, to test the change in covariance matrix at each time point. We study the asymptotic normality of Pillai’s statistics for testing the equality of two covariance matrices when both sample size and dimension increase at the same rate. We test the existence of a single change point in a given time period using Cauchy combination test, the test using an weighted sum of Cauchy transformed p-values, and estimate the change point as the point whose statistic is the greatest. To test and estimate multiple change points, we use the idea of the wild binary segmentation and repeatedly apply the procedure for a single change point to each segmented period until no significant change point exists. We numerically provide the size and power of our method. We finally apply our procedure to finding abnormal behavior in the investment of a private equity fund.

Change point detection in INAR(p) models via likelihood ratio scanning method

In this paper, we consider an integer-valued autoregressive (INAR) model. Based on the likelihood ratio scanning method (LRSM), the numbers and locations of the piecewise stationary INAR process change points are discussed. The minimum description length (MDL) function of the INAR model is given. Moreover, we also study the case when the sum of coefficients in each segment of the model tends to 1 and make adjustments to the MDL criterion in the LRSM. In this way, the accuracy of change point detection is improved. In the end, simulation experiments and real data analysis are conducted to illustrate the efficiency of the LRSM.

Change point detection in temporal networks based on graph snapshot similarity measures

This paper addresses the challenge of change point detection in temporal networks, a critical task across various domains, including life sciences and socioeconomic activities. Continuous analysis and problem-solving within dynamic networks are essential in these fields. While much attention has been given to binary cases, this study extends the scope to include change point detection in weighted networks, an important dimension of edge analysis in dynamic networks. We introduce a novel distance metric called the Interval Sum Absolute Difference Distance (ISADD) to measure the distance between two graph snapshots. Additionally, we apply a Gaussian radial basis function to transform this distance into a similarity score between graph snapshots. This similarity score function effectively identifies individual change points. Furthermore, we employ a bisection detection algorithm to extend the method to detect multiple change points. Experimental results on both simulated and real-world data demonstrate the efficacy of the proposed framework.

Advanced Bayesian Method for Timely Small-Scale Forest Loss Detection in the Brazilian Amazon and Cerrado with Sentinel-1 Time-Series

Abstract. The world’s forests are undergoing significant changes due to loss and degradation, emphasizing the need for Near Real-Time (NRT) monitoring to prevent further damage. Traditional monitoring methods using optical imagery are hindered by cloud coverage, while newer Synthetic Aperture Radar (SAR) systems, although operational in all weather conditions, face challenges such as sensitivity to soil moisture and the need for spatial filtering to reduce speckle effects. These limitations affect the detection of small-scale forest loss, especially in seasonally variable regions like dry forests and savannas. This paper presents a SAR-based forest disturbance detection method using Bayesian inference. Unlike traditional methods, this approach maintains the native resolution of the data by avoiding spatial filtering. Forest disturbance is modelled as a change-point detection problem within a non-filtered Sentinel-1 time series, where each new observation updates the probability of forest loss by leveraging prior information and a data model. This sequential adaptation ensures robustness against variations and trends, making it effective in monitoring disturbances across diverse forest types, including areas affected by seasonality. The proposed method was tested against other NRT monitoring systems for the year 2020, using small validation polygons (under 1 hectare) in the Brazilian Amazon and Cerrado savanna. Results demonstrate significant improvements in detecting small-scale disturbances and drastically reduced false alarm rates in both biomes. Notably, in the seasonality-sensitive Cerrado, our solution completely outperforms the leading and only existing optical technology.

CHANGEPOINTS DETECTION OF PANDEMIC WAVE IN REAL-TIME: APPLICATIONS TO THE TRANSMISSION OF COVID-19

The COVID-19 pandemic has spread throughout the world. Most countries experienced the pandemic in multiple waves. The Richards model predicts when a pandemic will peak and end in a particular area. However, this model can only be used in the single-wave case. The research aims to identify a changepoint detection method capable of delineating pandemic wave boundaries, thus enabling the resolution of multiple wave cases using the Richards model. This article uses two methods to detect changepoints: the Pruned Exact Linear Time (PELT) and the interpolation method. PELT method determines the changepoint based on changes in the statistical properties of the sequence of observations which can be in the form of differences in the mean or variance of each set of observations. In contrast, the linear interpolation method determines the changepoint based on the slope of a data pattern. The two methods complement each other, where the interpolation method is used to determine whether the pandemic is still in a single wave or has multiple waves, followed by determining wave boundaries using the PELT method. Richards model parameter estimation is carried out after the wave boundaries are obtained, and initial data is taken from the last wave using the PELT method. The prediction results show the peak of the pandemic in a particular region and when it will end, which can be used to inform medium-term strategies for the government to overcome the ongoing pandemic. This information helps prevent a resurgence of infections, which would negatively affect the COVID-19 mortality rate and the area's economic situation.

Monitoring event data extracted from online news for outbreak detection

Abstract Introduction Digital disease surveillance (DDS) detects public health events from internet-based data e.g., online news. Event features depicting epidemiological and social characteristics of health events can be extracted from news using the natural language process techniques. However, few studies have leveraged the event features to support anomaly detection in DDS. We aimed to understand the distribution of the event features and explore anomaly detection using the frequency of these features. Methods We collected event data from COVID-19-related news collected from October 1 to December 31, 2021, sourced from BioCaster, an infectious-disease-focused DDS system. The predefined event features in BioCaster include disease, pathogen, location and 14 binary features, such as if an event was caused by an unclassified virus. We described the distribution of the features and detected changes in the frequency of event features using a Bayesian online change point detection. We compared the change points with the number of new cases and of genomic samples collected. Results We included 170,168 news articles reporting COVID-19 in 155 countries. The event feature indicating that an event was caused by an unclassified virus was identified as positive among 3831 (2.25%) news and 12.91% of news had positive value for the feature indicating cases who had travelled across international borders. The change points detected from these two features were temporally correlated to the emergence of the Omicron variant in corresponding countries, which was more significant in countries with at least 300 news articles. Conversely, event features irrelevant to this case study, e.g., if the cases were military workers, were identified as negative in all news and no change points were detected. Conclusions Our study highlights the potential of monitoring the frequency of event features extracted from online news for anomaly detection in DDS, which relies on sufficient news coverage. Key messages • Monitoring the event features extracted from online news provide is useful approach for automatic anomaly detection in digital disease surveillance. • Increasing media coverage is fundamental for improving the early detection in a digital disease surveillance system.

Acquiring better load estimates by combining anomaly and change point detection in power grid time-series measurements

In this paper we present novel methodology for automatic anomaly and switch event filtering to improve load estimation in power grid systems. By leveraging unsupervised methods with supervised optimization, our approach prioritizes interpretability while ensuring robust and generalizable performance on unseen data. Through experimentation, a combination of binary segmentation for change point detection and statistical process control for anomaly detection emerges as the most effective strategy, specifically when ensembled in a novel sequential manner. Results indicate the clear wasted potential when filtering is not applied. The automatic load estimation is also fairly accurate, with approximately 90% of estimates falling within a 10% error margin, with only a single significant failure in both the minimum and maximum load estimates across 60 measurements in the test set. Our methodology’s interpretability makes it particularly suitable for critical infrastructure planning, thereby enhancing decision-making processes.

Multiscale Change Point Detection for Univariate Time Series Data with Missing Value

This paper studies the autoregressive integrated moving average (ARIMA) state space model combined with Kalman smoothing to impute missing values in a univariate time series before detecting change points. We estimate a scale-dependent time-average variance constant that depends on the length of the data section and is robust to mean shifts under serial dependence. The consistency of the proposed estimator is shown under the assumption allowing heavy tailedness. Integrating the proposed estimator with the moving sum and wild binary segmentation procedures to determine the number and locations of change points is discussed. Furthermore, the performance of the proposed methods is evaluated through extensive simulation studies and applied to the Beijing multi-site air quality dataset to impute missing values and detect mean changes in the data.

Feature selection based on game theory optimization to achieve desired performance metrics in seizure onset detection

Background and Objective: Epilepsy is one of the common diseases of the nervous system, which affects almost 1% of the world’s population. The unpredictable nature of epileptic seizures has caused people suffering from it to be deprived of their normal life, and making a device that can detect a seizure onset can help them. Methods:Game theory is a branch of mathematical science that, after proving its ability in economics, has entered into the field of signal processing and is trying to solve complex problems with a new approach. In this paper, we use game theory and coalitional game concepts to provide a novel way of feature selection in the analysis of Electroencephalography (EEG) signals. We have used a variety of statistical, morphological, and frequency features, and with the proposed feature selection algorithm we have attempted to select the smallest coalition which discriminates between two brain states while appropriately allocating resources to satisfy our desired sensitivity, delay, and false alarm. A modified change point detection algorithm has also been introduced to be used to detect epileptic seizures. Results:The results of the test on the CHB-MIT dataset show that using the proposed algorithm, we can achieve improved performance in terms of sensitivity, delay, and false alarm with a few number of features. Our framework also provides a means to focus on desired criteria for different applications. In scenario which the focus was on sensitivity, the proposed framework achieved 99% average sensitivity, 0.4/h average false alarm, and 2.4 s average delay with only using about 2 features for each patient. Conclusion:By applying the proposed framework on the CHB-MIT dataset we observe that we can achieve better performance compared to recent studies with fewer features in terms of sensitivity, delay, and false alarm metrics. Such an algorithm can be used for the purpose of real-time detection and in wearable devices.

Changepoint detection on daily home activity pattern: a sliced Poisson process method.

The problem of health and care of people is being revolutionized. An important component of that revolution is disease prevention and health improvement from home. A natural approach to the health problem is monitoring changes in people's behavior or activities. These changes can be indicators of potential health problems. However, due to a person's daily pattern, changes will be observed throughout each day, with, eg, an increase of events around meal times and fewer events during the night. We do not wish to detect such within-day changes but rather changes in the daily behavior pattern from one day to the next. To this end, we assume the set of event times within a given day as a single observation. We model this observation as the realization of an inhomogeneous Poisson process where the rate function can vary with the time of day. Then, we propose to detect changes in the sequence of inhomogeneous Poisson processes. This approach is appropriate for many phenomena, particularly for home activity data. Our methodology is evaluated on simulated data. Overall, our approach uses local change information to detect changes across days. At the same time, it allows us to visualize and interpret the results, changes, and trends over time, allowing the detection of potential health decline.

Change-point detection in anomalous-diffusion trajectories utilising machine-learning-based uncertainty estimates

Abstract When recording the movement of individual animals, cells or molecules one will often observe changes in their diffusive behaviour at certain points in time along their trajectory. In order to capture the different diffusive modes assembled in such heterogeneous trajectories it becomes necessary to segment them by determining these change-points. Such a change-point detection can be challenging for conventional statistical methods, especially when the changes are subtle. We here apply Bayesian Deep Learning to obtain point-wise estimates of not only the anomalous diffusion exponent but also the uncertainties in these predictions from a single anomalous diffusion trajectory generated according to four theoretical models of anomalous diffusion. We show that we are able to achieve an accuracy similar to single-mode (without change-points) predictions as well as a well calibrated uncertainty predictions of this accuracy. Additionally, we find that the predicted uncertainties feature interesting behaviour at the change-points leading us to examine the capabilities of these predictions for change-point detection. While the series of predicted uncertainties on their own are not sufficient to improve change-point detection, they do lead to a performance boost when applied in combination with the predicted anomalous diffusion exponents.

ID: 325692 Design of a Digitally-triggered Patient Follow-up Algorithm for Change Point Detection After SCS Implant

ID: 325692 Design of a Digitally-triggered Patient Follow-up Algorithm for Change Point Detection After SCS Implant

How have atmospheric components of the local water cycle changed around the abrupt climatic shift over France?

The significant increase in surface air temperature experienced by Western Europe over the last few decades has resulted in an abrupt warming in France, around 1987/1988 years. This climatic shift impacted hydrological cycle, particularly by reducing runoff in spring and summer. Evapotranspiration and precipitation have been identified as the main drivers of climatic water balance. But the impact of the 1987/1988 climatic shift on local water cycle over France has not been quantified yet. This study tries to assess the consequences of this rapid warming on the main climatic components of local water cycle. Climate variables linked to the water cycle extracted from a reanalysed observed climate database are analysed using robust Bayesian change points detection and mean comparison techniques. After the abrupt rise in surface air temperature and surface solar radiation, water demand increases significantly on almost the entire French territory in spring, summer and autumn. Our results show that, from March to May, the vegetation cover is able to respond to this increase by drawing from the soil water reservoirs. But in summer, most of the territory is facing a significant rise in water constraint (i.e. difference between potential and actual evapotranspiration), extending on the last decade over autumn. In spring and summer, the increase in potential evapotranspiration is the main driver of the intensification of water constraint. In the beginning of autumn, longer dry spell also plays a major role in the lengthening of periods of water constraint. This innovative study highlights the specific impact of a climatic shift on the main components of the atmospheric water balance at regional and local scale. As the observed changes in climate hazard linked to water cycle affect growth cycle of the majority of the vegetation covers and crops, this could lead to a worsening of hydric stress events. As such climatic shifts are expected to happen again in the future, their impacts on the local water cycle represent a major issue for natural terrestrial ecosystems as well as for agriculture.

APFC: Adaptive Particle Filter for Change Point Detection of Profile Data in Manufacturing Systems

APFC: Adaptive Particle Filter for Change Point Detection of Profile Data in Manufacturing Systems

Exploitation of long-range acoustic dependence for early change detection in dynamic machine running status

This paper suggests that the long-range dependence (LRD) in acoustic signal is an important indicator of dynamic behavior in data. Considering this information, a novel framework based on LRD in acoustic is proposed for detecting change-points in machine running status. Initially, the LRD phenomenon in acoustic is demonstrated. Subsequently, a change-point detection framework is developed, leveraging LRD to monitor machine running states. The framework includes two phases: (1) characteristic data prediction: the LRD value is predicted by the fractional autoregressive integrated moving average (FARIMA) model, which is applied to reflect changes in machine operation states. (2) date change detection: based on residual analysis, a null hypothesis testing based automatic analysis method for acoustic signals during machine successive operations is used to make decision. Experiments in real-world applications demonstrated the good potential of the proposed framework in practical engineering scenarios.

TRACER Perspectives on Gulf-Breeze and Bay-Breeze Circulations and Coastal Convection

Abstract This study explores gulf-breeze circulations (GBCs) and bay-breeze circulations (BBCs) in Houston–Galveston, investigating their characteristics, large-scale weather influences, and impacts on surface properties, boundary layer updrafts, and convective clouds. The results are derived from a combination of datasets, including satellite observations, ground-based measurements, and reanalysis datasets, using machine learning, changepoint detection method, and Lagrangian cell tracking. We find that anticyclonic synoptic patterns during the summer months (June–September) favor GBC/BBC formation and the associated convective cloud development, representing 74% of cases. The main Tracking Aerosol Convection Interactions Experiment (TRACER) site located close to the Galveston Bay is influenced by both GBC and BBC, with nearly half of the cases showing evident BBC features. The site experiences early frontal passages ranging from 1040 to 1630 local time (LT), with 1300 LT being the most frequent. These fronts are stronger than those observed at the ancillary site which is located further inland from the Galveston Bay, including larger changes in surface temperature, moisture, and wind speed. Furthermore, these fronts trigger boundary layer updrafts, likely promoting isolated convective precipitating cores that are short lived (average convective lifetime of 63 min) and slow moving (average propagation speed of 5 m s−1), primarily within 20–40 km from the coast.

Unveiling change-point detection with ARMA-GARCH models

This paper delves into the asymptotic theory of a Cumulative Sum (CUSUM)-type test designed for detecting change points in the variance of AutoRegressive Moving-Average models, incorporating Generalized Autoregressive Conditional Heteroscedasticity innovations. We demonstrate that, under the null hypothesis (no change), the CUSUM test statistic converges to the supremum of a standard Brownian bridge. Using a Monte Carlo simulation, we highlight the strong performance of our proposed test when compared to the methods by Song and Kang. Furthermore, we provide a real-world data analysis to showcase the practical application of our test.

Climate Extremes Related with Temperature Change Point Detection in Pambar Kottakaraiyar Basin

Global climate change has changed the trend of climate variables. Also there is change in intensity, frequency and cyclic pattern of climate extremes such as flood, drought, heatwaves and tropical cyclones in the southern part of India. The high temperatures and further increase pose serious socio-economic and environmental impacts. There is an urgent necessity to know about the regional level climate extremes and their adaptation strategies especially for the drought prone and rain fed basins. This paper discusses about the temperature trend and change point over a long term historical time period of 30 years (1992-2021). Based on the Man-Kendall method and Sen’s test, the temperature over the basin is showing a significant rising trend with positive slopes in almost all the months, seasons and annual statistics. The change points of mean temperature are observed in 2008 for both Kundrakudi and Pudukottai climate stations. The minimum temperature change point was identified in 1998 and 2009 years for the Kundrakudi climate station. The results of this paper will be helpful in capacity building of specialised team on extremes risk, vulnerability and sector specific indices for temperature for the water resources and agricultural sectors under changing climate.

Research on Change Point Detection during Periods of Sharp Fluctuations in Stock Prices–Based on Bayes Method β-ARCH Models

Research on Change Point Detection during Periods of Sharp Fluctuations in Stock Prices–Based on Bayes Method β-ARCH Models