Mutual Information-Based Variable Selection on Latent Class Cluster Analysis

Objective: The cognition of Alzheimer's disease (AD) has a heterogeneous pattern. It is useful to obtain more information about specific subgroups of patients to prevent disease progression. For better identification of the population, we aimed to detect latent groups based on cognitive test scores using latent class (LC) cluster analysis and influencing factors of latent severity groups to assist practitioners in outpatient departments who have restricted time and instrumentation. Materials and Methods: Data for 630 patients with AD in the Mersin University Dementia Outpatient Unit were collected, and cognitive test scores, demographic variables, and other factors such as comorbidities and family history of dementia were obtained. Initially, LC cluster analysis was performed to distinguish subgroups considering clinical dementia scores, age, and sex as covariates. Second, univariate analysis was used to detect the relationship between latent subgroups and influencing factors. Finally, multinomial logistic regression was performed to identify the magnitude of risk for significant factors. Results: Four severity groups were defined as mild, moderate, severe, and very severe cases of AD, and severity was significantly related to educational level, hyperlipidemia, diabetes mellitus, and sarcopenia (P < 0.001, P = 0.001, P = 0.043, and P < 0.001, respectively). Family history also influenced severity (P = 0.024). Disease severity increased with decreased education levels. Family history predicted a 1.555-fold increase in the risk of being in the moderate group versus the mild group. Moreover, diabetes mellitus predicted a 3.690-fold increase of being in the very severe group versus the mild group. Conclusion: LC cluster analysis is effective for determining severity groups for AD, and study results will help prepare a guide for an optimum evaluation tool for the disease.

Mixed integer second-order cone programming formulations for variable selection in linear regression

Abstract. Among the natural hazards, flash flooding is the leading cause of weather-related deaths. Flood risk management (FRM) in this context requires a comprehensive assessment of the social risk component. In this regard, integrated social vulnerability (ISV) can incorporate spatial distribution and contribution and the combined effect of exposure, sensitivity and resilience to total vulnerability, although these components are often disregarded. ISV is defined by the demographic and socio-economic characteristics that condition a population's capacity to cope with, resist and recover from risk and can be expressed as the integrated social vulnerability index (ISVI). This study describes a methodological approach towards constructing the ISVI in urban areas prone to flash flooding in Castilla y León (Castile and León, northern central Spain, 94 223 km2, 2 478 376 inhabitants). A hierarchical segmentation analysis (HSA) was performed prior to the principal components analysis (PCA), which helped to overcome the sample size limitation inherent in PCA. ISVI was obtained from weighting vulnerability factors based on the tolerance statistic. In addition, latent class cluster analysis (LCCA) was carried out to identify spatial patterns of vulnerability within the study area. Our results show that the ISVI has high spatial variability. Moreover, the source of vulnerability in each urban area cluster can be identified from LCCA. These findings make it possible to design tailor-made strategies for FRM, thereby increasing the efficiency of plans and policies and helping to reduce the cost of mitigation measures.

Exploring bicyclist injury severity in bicycle-vehicle crashes using latent class clustering analysis and partial proportional odds models

The market share of electric vehicles (EVs) is growing rapidly, making it crucial to understand the charging behaviour of current and prospective users. Such understanding is essential for designing policies that positively influence consumers’ charging behaviour and facilitate EV adoption. In this study, we examined the heterogeneity in charging preferences of 994 respondents across Australia using a latent class cluster model that considers indicators of charging behaviour as outcomes of interest. We used sociodemographic characteristics, travel needs, and EV adoption status as covariates to predict class membership. Our findings indicate five segments of consumers with distinct charging preferences: routine-focused frugals, cost-oriented deliberators, range seekers, flexibility seekers, and indifferent late adopters. These segments differ in the importance they attach to charging attributes, their coping strategies with limited battery resources, and their risk attitude. Our results suggest that a uniform approach to EV-related policies is not appropriate, as each consumer segment has unique charging preferences and requirements. Furthermore, the study emphasizes the significance of accounting for charging behaviour heterogeneity in demand modelling, as assumptions in current models may not accurately represent the decision-making of most segments.

Symptom burden associated with chronic kidney disease can be debilitating, with a negative effect on patient health-related quality of life. Latent class clustering analysis is an innovative tool for classifying patient symptom experience. The aim of the study was to identify subgroups of patients at greatest risk for high symptom burden, which may facilitate development of patient-centered symptom management interventions. In this cross-sectional analysis, baseline data were analyzed from 3,921 adults enrolled in the Chronic Renal Insufficiency Cohort Study from 2003 to 2008. Latent class cluster modeling using 11 items on the Kidney Disease Quality of Life symptom profile was employed to identify patient subgroups based on similar observed physical symptom response patterns. Multinomial logistic regression models were estimated with demographic variables, lifestyle and clinical variables, and self-reported measures (Kidney Disease Quality of Life physical and mental component summaries and the Beck Depression Inventory). Three symptom-based subgroups were identified, differing in severity (low symptom, moderate symptom, and high symptom). After adjusting for other variables in multinomial logistic regression, membership in the high-symptom subgroup was less likely for non-Hispanic Blacks and men. Other factors associated with membership in the high-symptom subgroup included lower estimated glomerular filtration rate, history of cardiac/cardiovascular disease, higher Beck Depression Inventory scores, and lower Kidney Disease Quality of Life physical and mental component summaries. Three symptom subgroups of patients were identified among patients with mild-to-moderate chronic kidney disease. Several demographic and clinical variables predicted membership in subgroups. Further research is needed to determine if symptom subgroups are stable over time and can be used to predict healthcare utilization and clinical outcomes.

The problem of variable selection is one of the most pervasive model selection problems in statistical applications. Often referred to as the problem of subset selection, it arises when one wants to model the relationship between a variable of interest and a subset of potential explanatory variables or predictors, but there is uncertainty about which subset to use. Several papers have dealt with various aspects of the problem but it appears that the typical regression user has not benefited appreciably. One reason for the lack of resolution of the problem is the fact that it is has not been well defined. Indeed, it is apparent that there is not a single problem, but rather several problems for which different answers might be appropriate. The intent of this chapter is not to give specific answers but merely to present a new simple multiplicative variable selection criterion based on the parametrically penalized residual sum of squares to address the subset selection problem in multiple linear regression analysis, where the objective is to select a minimal subset of predictor variables without sacrificing any explanatory power. The variables, which optimize this criterion, are chosen to be the best variables. The authors find that the proposed criterion performs consistently well across a wide variety of variable selection problems. Practical utility of this criterion is demonstrated by numerical examples.

Subset selection by Mallows’ [formula omitted]: A mixed integer programming approach

Digital technologies make significant impact on labor market, primarily by complementing or by substituting workers. This has triggered a change in the set of skills that are required from workers, by putting stronger focus on problemsolving skills, creativity, socioemotional skills, functional literacy and technical skills related to the use of digital technologies. The effects of digitalization on the labor market and economic performances of a particular country in the future depend on the workforce adaptability, industrial and occupational structure, the skills mix, organization of work and current state of digitalization. The aim of this paper is to evaluate the degree of workforce skills adaptability in 30 European countries, using the OECD data on achievement in reading, math and science, as well as the data on digital competencies, inclusion in lifelong learning and subjective perception on ability to find a new job. Our results suggest positive relationship between adaptability and PISA results. Using the principal component analysis, cluster analysis and LCCA (latent class cluster analysis), we find that European countries can be grouped into three clusters, in terms of adaptability: high performing (North and Western Europe), medium performing (Central Europe and Baltics) and low performing (South and South-eastern Europe). For some countries, low levels of adaptability of the workforce can pose an important obstacle for future growth and development.

Survey data on forest owner attitudes can be used to divide forest owners into types, each type forming a group of similar owners with attitudes differing markedly from those observed for other types. Such an empirical typology can be developed using a range of different statistical methods, including cluster analyses and latent class analysis. In this paper we examine the sensitivity of a forest owner typology to the choice of statistical model. Based on a latent class analysis of data from a survey among private forest owners in Denmark, we identify different types of owners. The resulting typology is compared with previous results based on k-means clustering of the same data. It emerges that the two statistical methods lead to almost identical forest owner types, but also that the final typology is influences by various choices made on the way through the analysis. The most significant of these choices is the number of owner types.

NEO-FFI personality clusters in trichotillomania

Do performance goals promote learning? A pattern analysis of Singapore students’ achievement goals

This study assessed the knowledge of heart attack symptoms and risk factors in a convenience sample of Korean immigrants. A total of 116 Korean immigrants in a Midwestern metropolitan area were recruited through Korean churches and markets. Knowledge was assessed using both open-ended questions and a structured questionnaire. Latent class cluster analysis and Chi-square tests were used to analyze the data. About 76% of the sample had at least one self-reported risk factor for cardiovascular disease. Using an open-ended question, the majority of subjects could only identify one symptom. In the structured questionnaire, subjects identified a mean of 5 out of 10 heart attack symptoms and a mean of 5 out of 9 heart attack risk factors. Latent class cluster analysis showed that subjects clustered into two groups for both risk factors and symptoms: a high knowledge group and a low knowledge group. Subjects who clustered into the risk factor low knowledge group (48%) were more likely than the risk factor high knowledge group to be older than 65 years, to have lower education, to not know to use 911 when a heart attack occurred, and to not have a family history of heart attack. Korean immigrants' knowledge of heart attack symptoms and risk factors was variable, ranging from high to very low. Education should be focused on those at highest risk for a heart attack, which includes the elderly and those with risk factors.

Identifying e-scooter rider profiles in the United States: A latent class cluster analysis

Identifying Daily and Weekly Charging Profiles of Electric Vehicle Users in Korea : An Application of Sequence Analysis and Latent Class Cluster Analysis