Topic
Change-Point Analysis
About: Change-Point Analysis is a research topic. Over the lifetime, 18 publications have been published within this topic receiving 371 citations.
Papers
1 Jan 1994
TL;DR: A brief account of the areas of change point analysis can be found in this paper, where the most basic problems are those of testing the hypothesis of uno change and estimating a change point by a point estimator or a confidence set when the presence of one is suspected.
Abstract: Change-points divide statistical models into homogeneous segments. Inference about change-points is discussed here in the context of testing the hypothesis of 'no change', point and interval estimation of a change-point, changes in nonparametric models, changes in regression, and detection of change in distribution of sequentially observed data. 1? Introduction. Suppose that in a linear array of independent observations Y\,.. .,Yn, the distribution is subject to change after Yr for some 1 < t < ? - 1. Detection and estimation of change-points which in this way divide statistical models into homogeneous segments is a fast-developing area of research in statistical theory and methods. We shall present here a brief account of some of the areas of change-point analysis. The most basic problems are those of testing the hypothesis of uno change," and of estimating a change-point by a point estimator or a confidence set when the presence of one is suspected. In Sections 2-4, we shall discuss these problems and some nonparametric methods will be presented in Section 5. Change-point problems also occur in the context of regression when the nature of dependence of one variate on another may be different in two segments of the data, and in situations where the observations are obtained sequentially with the possibility of a change in distribution at any stage. Methods in these two areas will be discussed in Sections 6 and 7.
134 citations
TL;DR: This paper provides a methodology to test existence, type, and strength of changes in the distribution of a sequence of hydrometeorological random variables based on Bayesian model selection and is illustrated using univariate normal models.
107 citations
TL;DR: In this paper, the authors revisited the annual mean rainfall data from Tucuman, Argentina for the years 1884-1996 for an in-depth change-point analysis and showed that change detection statistics are sensitive to how the model variance is estimated, and even marginally significant serial correlations among the observations can have a highly significant effect upon the variance estimate.
Abstract: The annual mean rainfall data from Tucuman, Argentina for the years 1884–1996 is revisited for an in-depth change-point analysis Applying classical change detection statistics, we first demonstrate that change-point analysis under independence is invalid Upon properly adjusting for dependence, it is found that a significant change has occurred in the mean of the data in and around 1956 The data analysis demonstrates that change detection statistics are sensitive to how the model variance is estimated Moreover, the analysis shows that even marginally significant serial correlations among the observations can have a highly significant effect upon the variance estimate Finally, the analysis illustrates that one should exercise care while implementing parametric or non-parametric methods in estimating the model variance, and one should be prepared for one of the methods outperforming the other Copyright © 2010 John Wiley & Sons, Ltd
17 citations
TL;DR: In this paper, the authors aim at estimating the multiple change points for the time series data of COVID-19 confirmed cases and deaths and trend estimation within the estimated MCP in India as compared with WHO regions.
Abstract: The present study aims at estimating the multiple change points for the time series data of COVID-19 confirmed cases and deaths and trend estimation within the estimated multiple change points (MCP) in India as compared with WHO regions. The data were described using descriptive statistical measures, and for the estimation of change point's E-divisive procedure was employed. Further, the trend within the estimated change points was tested using Sen's slope and Mann Kendal tests. India, along with the African Region, American region, and South East Asia regions experienced a significant surge in the fresh cases up to the 5th Change point. Among the WHO regions, The American region was the worst hit by the pandemic in case of fresh cases and deaths. While the European region experienced an early negative trend of fresh cases during the 3rd and 4th change point, but later the situation reversed by the 5th (7th July 2020) and 6th (6th August 2020) change point. The trend of deaths in India and the South-East Asia Region was similar, and global deaths had a negative trend from the 4th (17th May 2020) Change point onwards. The change points were estimated with prefixed significance level α < 0.002. Infections and deaths were positively significant for India and SEARO region across change points. Infection was significant at every 30 days interval across other WHO regions, and any delay in the infections was due to the interventions. The European region is expected to have a second wave of positive infections during the 5th and 6th change points though the early two change points were negatively significant. The study highlights the efficacy of change point analysis in understanding the dynamics of covid-19 cases in India and across the world. It further helps to develop effective public health strategies.
15 citations
Related Topics (5)
Performance Metrics
| Year | Papers |
|---|---|
| 2021 | 1 |
| 2020 | 1 |
| 2018 | 2 |
| 2017 | 1 |
| 2014 | 2 |
| 2013 | 1 |