Article
Mathematics
Persi Diaconis
Summary: A parameter free version of classical models for contingency tables is developed based on de Finetti's notions of partial exchangeability.
Article
Computer Science, Artificial Intelligence
Yan Xia, Nishant Ravikumar, Alejandro F. Frangi
Summary: This study proposes a two-stage pipeline for detecting and synthesizing missing slices in cardiac MR acquisition. The proposed network can infer anatomically plausible missing slices and improve the accuracy of subsequent analyses on cardiac MRIs through synthesized image slices. Experimental results show that this method outperforms incomplete data in analysis.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Multidisciplinary Sciences
Kengo Fujisawa, Kouji Tahata
Summary: This paper proposes an asymmetry plus association model for analyzing square contingency tables with ordinal categories, and partitions the test statistic for goodness-of-fit using the proposed model.
Article
Statistics & Probability
Zheng Wei, Daeyoung Kim, Erin M. Conlon
Summary: A fully Bayesian method was developed to implement a subcopula-based asymmetric association measure for variables in two-way contingency tables, and its performance was examined using simulation data and real data sets of colorectal cancer. The Bayesian method outperformed the large-sample method on average in simulation studies, and provided comparable or improved results for the real data analyses.
COMPUTATIONAL STATISTICS
(2022)
Article
Computer Science, Artificial Intelligence
Lang-wangqing Suo, Hai-Long Yang
Summary: This paper presents a preliminary approach to three-way conflict analysis models based on incomplete situation tables, constructing models based on different variants of incomplete situation tables and demonstrating the relationships of conflict, neutrality, and alliance among agents using examples.
INTERNATIONAL JOURNAL OF APPROXIMATE REASONING
(2022)
Article
Statistics & Probability
Ruiping Liu, Ndeye Niang, Gilbert Saporta, Huiwen Wang
Summary: We propose sparse variants of correspondence analysis (CA) for large contingency tables in text mining. Sparse CA seeks to obtain zero coefficients to address the difficulty in interpreting CA results when the table size is large. We adapt known sparse versions of weighted PCA and generalized SVD, and develop specific methods for obtaining orthogonal solutions and tuning sparseness parameters. We distinguish between cases where sparseness is applied to both rows and columns, or only one set.
ADVANCES IN DATA ANALYSIS AND CLASSIFICATION
(2023)
Article
Multidisciplinary Sciences
Ken Saito, Nozomi Takakubo, Aki Ishii, Tomoyuki Nakagawa, Sadao Tomizawa
Summary: In this article, various types of marginal homogeneity models are proposed to determine changes in political party support. The authors introduce local marginal homogeneity models and two measures to express the degree of departure from the models. The measures are applied to data and prove to be effective in comparing the degree of departure in different tables.
Article
Computer Science, Artificial Intelligence
Jan Flusser, Tomas Suk, Barbara Zitova
Summary: The paper demonstrates that the moment invariants proposed by Hjouji et al. are incomplete, leading to limited discriminability. This is proven through circular projection of the image, and in a broader context, completeness of invariants is shown to enhance recognition power.
JOURNAL OF MATHEMATICAL IMAGING AND VISION
(2021)
Article
Multidisciplinary Sciences
Yusuke Saigusa, Yuta Teramoto, Sadao Tomizawa
Summary: A novel measure is proposed in this paper for the analysis of square contingency tables with ordered categories, based on the departure from conditional symmetry using cumulative probabilities from the corners of the square table. This measure is applied to Japanese occupational status data to illustrate the departure from a proportional structure of social mobility.
Article
Computer Science, Interdisciplinary Applications
Nicole S. Erler, Dimitris Rizopoulos, Emmanuel M. E. H. Lesaffre
Summary: Missing data is a common issue in studies and can complicate analysis. While multiple imputation methods are popular and work well in standard settings, they may result in bias in settings involving nonlinear associations or interactions. Additionally, complex outcomes like longitudinal or survival data are not adequately addressed by standard implementations. The R package JointAI aims to overcome these issues by utilizing a Bayesian framework for simultaneous analysis and imputation in regression models with incomplete covariates, providing flexibility and compatibility with various types of data.
JOURNAL OF STATISTICAL SOFTWARE
(2021)
Article
Computer Science, Information Systems
Xiaoxue Liu, Jiexin Zhang, Peidong Zhu, Qingping Tan, Wei Yin
Summary: This paper presents a unified methodology for quantitatively and automatically analyzing cyber-physical attacks on ICSs. By defining the weighted colored Petri net and basic cyber-physical attack models, as well as proposing a method to calculate weights in attack models, the study shows stable weights and establishes threat propagation matrix and security state vector. Additionally, a cyber-physical attack path analysis algorithm is designed to discover possible attack paths with specific attack losses.
COMPUTERS & SECURITY
(2021)
Article
Statistics & Probability
Shiva S. Dibaj, Alan D. Hutson, Graham W. Warren, Gregory E. Wilding
Summary: With recent developments in computer power, the application of exact inferential methods has become more popular. However, there is a lack of such methodology for populations with complex structures, such as finite populations. To address this issue, the researchers developed an exact unconditional test for comparing proportions in finite populations. The proposed test utilizes information from the sample to restrict the search for the maximum p-value, while maintaining the specified nominal significance level.
JOURNAL OF APPLIED STATISTICS
(2022)
Article
Oncology
Felix Behling, Julia Rang, Elena Dangel, Susan Noell, Mirjam Renovanz, Irina Maeurer, Jens Schittenhelm, Benjamin Bender, Frank Paulsen, Bettina Brendel, Peter Martus, Jens Gempt, Melanie Barz, Bernhard Meyer, Marcos Tatagiba, Marco Skardelly
Summary: Complete and incomplete recurrent resection are beneficial for post-progression survival in progressive GBM patients, regardless of MGMT methylation, age, or adjuvant therapy, but not for patients with a poor clinical condition with a KPS <70.
FRONTIERS IN ONCOLOGY
(2022)
Article
Computer Science, Information Systems
Shuping Zhao, Lunke Fei, Jie Wen, Jigang Wu, Bob Zhang
Summary: In the real-world, missing views are a common problem in multiview data. Existing clustering methods often ignore the hidden information of the absent views and fail to reflect the real distribution of the data. To address this issue, this paper proposes a method called ICSL_IMC, which learns the intrinsic and complete structures for all views, effectively capturing the real distribution of the absent instances. The proposed method outperforms state-of-the-art methods on various databases.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Mathematical & Computational Biology
Marina Vives-Mestres, Amparo Casanova
Summary: Two-way contingency tables are commonly seen in medical studies, and can be analyzed and visualized using CoDa methods. The relationships, strength and direction of dependency between variables can be represented by logratios in a quaternary diagram. Additionally, a simplicial regression model can be used to model the tables and illustrate prediction abilities.
STATISTICS IN MEDICINE
(2021)
