Polychotomous Regression Model Research Articles

Regresi Logistik Ordinal untuk Memodelkan Predikat Lulusan Perguruan Tingggi

Logistic regression is an alternative model that can model the relationship between a categorical response variable and one or more categorical, continuous predictor variables, or a combination of categorical and continuous predictor variables. Based on the number of categories in the response variable, the logistic regression model consists of a dichotomous logistic regression model and a polychotomous regression model. The dichotomous logistic regression model is a logistic regression model that has two categories in the response variable and has a Bernoulli distribution. In comparison, the polychotomous logistic regression model is a logistic regression model that has three or more categories and a multinomial distribution. The polychotomous logistic regression model is divided into two models, namely the multinomial logistic regression model and ordinal logistic regression. This research aims to examine ordinal logistic regression modeling and its application to the predicate of graduates of the undergraduate program at the Faculty of Mathematics and Natural Sciences, Mulawarman University (FMIPA UNMUL) for the 2020 graduation period. The results of the research show that the factors that have a significant influence on the predicate of graduates of the FMIPA UNMUL undergraduate program are gender and admission route. Female graduates of the FMIPA UNMUL undergraduate program have a greater chance of achieving satisfactory and very satisfactory predicates compared to achieving a cum laude predicate. Graduates of the FMIPA UNMUL undergraduate program who are accepted through the SMMPTN admission route have a lower chance of achieving satisfactory and very satisfactory predicates compared to achieving a cum laude predicate.

Firm-Level Investment Decisions Under Uncertainty and Irreversibility in Zimbabwes Private Firms

Zimbabwe’s business operating environment is abounding with both micro and macro uncertainties that regularly impact on firm-level investment decisions. Under uncertainty and irreversibility private firms are forced to integrate into their investment appraisals and decisions some expectations regarding future product demand and prices, likely returns on fixed capital investment, expected margins, cost of capital expectations and cash flow projections. These important issues are not known with definitive assurance and hence, making investment decisions under uncertainty a challenging task for Zimbabwe’s private firms. The study examined investment under uncertainty and irreversibility in Zimbabwe’s private firms using a polychotomous regression model with the three investment decision outcomes; “invest now”, “do not invest now” and “defer investment”. The major findings are that; (1) liquidity constraints, firm size and credit constraints significantly influence the probability of investing now irrespective of uncertainties in the economy. (2) Poor public infrastructure, political uncertainty, inconsistent application of the indigenization laws and absence of laws that protect private property rights increase the probability of private firms’ deferring investment decisions under uncertainty and investment irreversibility. The paper recommends that policy makers should reduce macro uncertainties in the financial sector that affect firm-level investment decisions, adopt policies that enhance productivity of public infrastructure and must observe national and international laws that safeguard investor property rights.

Does Racial Variation Influence Preoperative Characteristics and Intraoperative Findings in Patients Undergoing Anterior Cruciate Ligament Reconstruction?

Background: A limited number of large multiethnic cohorts of patients undergoing anterior cruciate ligament reconstruction (ACLR) exist; therefore, little is known about racial differences in preoperative and intraoperative characteristics of patients undergoing this procedure. Purpose: To evaluate preoperative patient, surgeon, and hospital characteristics and intraoperative findings associated with various patient races undergoing ACLR. Study Design: Cohort study; Level of evidence, 3. Methods: A cross-sectional study of patients undergoing primary ACLR (performed between January 1, 2008 and September 30, 2012) was conducted. The Kaiser Permanente ACLR Registry was used to identify the cases and variables for the study. The variables of age, sex, time from injury to surgery, body mass index, sport at the time of injury, concomitant injury (menisci, cartilage, other ligament), surgeon training, and surgeon and site volumes were evaluated. Race was categorized into white (reference), Hispanic, Asian, and black. Polychotomous regression models were employed. Odds ratios (ORs) and 95% CIs are provided. Results: A total of 14,278 ACLRs were evaluated; there were 7401 (51.8%) ACLRs performed on white, 3912 (27.4%) on Hispanic, 1894 (13.3%) on Asian, and 1071 (7.5%) on black patients. Men predominated overall as well as in each of the race categories. Compared with white patients, after adjusting for all covariates, Asian (OR, 1.24; 95% CI, 1.10-1.40) and Hispanic (OR, 1.52; 95% CI, 1.39-1.67) patients undergoing ACLR were more likely to be male than female, and black patients were more likely to be female (OR, 0.69; 95% CI, 0.60-0.80). All racial groups tended to undergo ACLR in younger age bands compared with white patients, with black patients having the highest odds of being <17 years (OR, 2.74; 95% CI, 2.23-3.37) and 17 to 24 years of age (OR, 2.28; 95% CI, 1.94-2.68) compared with being ≥25 years. Asian patients were more likely injured playing basketball (OR, 4.31; 95% CI, 3.67-5.06), black patients during football (OR, 3.48; 95% CI, 2.73-4.45) and basketball (OR, 5.52; 95% CI, 4.49-6.78), and Hispanic patients during soccer (OR, 3.09; 95% CI, 2.75-3.48). Hispanic patients had a higher likelihood of having both menisci injured at the time of surgery (OR, 1.31; 95% CI, 1.16-1.49) as well as the lateral meniscus (OR, 1.31; 95% CI, 1.17-1.47). Asian patients had a lower likelihood of cartilage injuries (OR, 0.78; 95% CI, 0.68-0.88). Sports medicine fellowship–trained surgeons were more likely to have operated on black (OR, 1.27; 95% CI, 1.07-1.51) and Hispanic (OR, 1.16; 95% CI, 1.04-1.29) patients. Conclusion: In this large representative sample of patients undergoing ACLR in the United States, 48.2% of the cohort was nonwhite, and racial variations in sex, age, sport at injury, and intraoperative findings at ACLR were identified. Understanding the differences in ACL injury presentation and concomitant injuries by race can be useful in helping tailor the development of an ACL injury prevention program, during the informed consent process, and in the arthroscopic portion of the procedure according to patient race–specific characteristics.

Dynamic Latent Class Model Averaging for Online Prediction

ABSTRACTWe consider the problem of online prediction when it is uncertain what the best prediction model to use is. We develop a method called dynamic latent class model averaging, which combines a state‐space model for the parameters of each of the candidate models of the system with a Markov chain model for the best model. We propose a polychotomous regression model for the transition weights to assume that the probability of a change in time depends on the past through the values of the most recent time periods and spatial correlation among the regions. The evolution of the parameters in each submodel is defined by exponential forgetting. This structure allows the ‘correct’ model to vary over both time and regions. In contrast to existing methods, the proposed model naturally incorporates clustering and prediction analysis in a single unified framework. We develop an efficient Gibbs algorithm for computation, and we demonstrate the value of our framework on simulated experiments and on a real‐world problem: forecasting IBM's corporate revenue. Copyright © 2014 John Wiley &amp; Sons, Ltd.

연속형의 텐서곱과 범주형의 직합을 사용한 다항 로지스틱 회귀모형

다항 로지스틱 회귀모형의 설명변수가 연속형과 범주형을 모두 포함할 때 범주형 설명변수는 직합을 적용하고 연속형 설명변수는 텐서곱을 적용하는 모형을 제안한다. 변수선택의 기준으로 BIC를 사용하고, 제안된 모형의 알고리즘을 구현하였다. 구현된 알고리즘을 실제 자료에 적용하여 기존의 방법과 비교하여 제안된 모형이 더 좋은 분류율을 보임을 확인하였다. In this paper, we propose a polychotomous regression model when independent variables include both categorical and numerical variables. For categorical independent variables, we use direct sums, and tensor product splines are used for continuous independent variables. We use BIC for varible selections criterior. We implemented the algorithm and apply the algorithm to real data. The use of direct sums and tensor products outperformed the usual multinomial logistic regression model.

Patient and Surgeon Characteristics Associated With Primary Anterior Cruciate Ligament Reconstruction Graft Selection

Background: It has been suggested that a surgeon’s experience and training are the most important factors associated with graft selection, but no studies have qualified this association. Graft usage prevalence has not been described for large anterior cruciate ligament reconstruction (ACLR) populations in the United States. Purpose: To describe the prevalence of graft usage in a large community-based practice and evaluate the association of patient, surgeon, and site characteristics with choice of primary ACLR graft. Study Design: Cross-sectional study; Level of evidence, 3. Methods: Primary ACLRs performed between February 2005 and June 2010 were selected for the study. A community-based ligament registry was used to identify cases and variables used for analysis. Graft choice (any allograft, hamstring autograft, and bone–patellar tendon–bone [BPTB] autograft) was compared by patient characteristics and surgeon and site characteristics. Associations between independent variables and graft choice were evaluated using a polychotomous regression model. Results: Of the 9849 patients included in the study, 64% were male, and overall median age was 28 years. Of these, 2796 (28.4%) received BPTB autografts, 3013 (30.6%) received hamstring autografts, and 4040 (41.0%) received allografts. The prevalence of graft source by patients’ gender, race, age, body mass index (BMI), as well as surgeons’ fellowship training status, average volume, and site volume were significantly different (all P < .001). Adjusted models showed that patients’ gender (P < .001), race (P = .018), age (P < .001), BMI (P < .001), as well as surgeons’ fellowship training status (P < .001), average volume (P < .001), and site volume (P < .001) are associated with graft selection. Older and female patients with lower BMI were more likely to receive allografts and hamstring autografts than BPTB autografts. Cases performed by non–fellowship-trained surgeons, lower volume sites, and/or lower volume surgeons were also more likely to be performed with allografts or hamstring autografts than BPTB autografts. Conclusion: Gender, age, race, as well as facility and surgeon characteristics such as volume and location are associated with ACL graft choices.

Do christian denominations exhibit higher rates of alcohol consumption? A study of Korean American women in California.

Although Korean American women show high levels of involvement in religious practices and high prevalence of alcohol consumption, no studies have assessed the association between religious denomination and alcohol intake among this group of women. This cross-sectional study examined the associations of religious denomination and religious commitment to alcohol consumption among Korean American women in California. Polychotomous regression models were used to provide estimates of the associations between religious denomination and religious commitment to alcohol consumption. Catholic Korean American women (OR 5.61 P < 0.01) and Independent Christian women (OR 4.87 P < 0.01) showed stronger associations to heavy alcohol consumption when compared to Conservative Christian Korean American women. Path analysis suggested that specific denominations had both direct and indirect effects on the outcome of interest, and that religious commitment and drinking models served as moderators for this phenomenon.

An Improved Goodness of Fit Statistic for Probability Prediction Models

We consider the general case of probability prediction models having two or more outcomes and propose an adjusted χ2 statistic which can be used to assess the goodness of fit of these models. We present a simulation study to show that our proposed statistic has an approximate χ2 distribution under the null hypothesis. Two applications are provided to illustrate the use of the new statistic. The first application examines the fit of a logistic regression model using both the proposed statistic and the popular Hosmer-Lemeshow statistic and we compare and contrast these two methods. The second application evaluates the goodness of fit of a polychotomous regression model.

The trading of sex for money or drugs, sexually transmitted diseases (STDs), and HIV-related risk behaviors among multisubstance using alcoholic inpatients

Data from 802 multisubstance using alcoholic inpatients (481 men and 321 women) were analyzed to determine whether trading sex for money or drugs was associated with self-reported STDs, self-reported HIV infection, and perceptions of current and future probable infection. Logistic and ordinal polychotomous regression models were used to evaluate the statistical significance of sex trade on STDs and HIV-related behaviors while controlling for other potential confounding variables such as cocaine abuse and IV drug use. Involvement in sex trade was associated with higher cocaine abuse, with STDs, and with self-reported HIV infection. The multivariate (adjusted) models supported the significance of sex trade with regard to STDs and HIV-related behaviors. The findings indicated that, among these multisubstance using alcoholic inpatients, high levels of trading sex for money or drugs were independently associated with self-reported STDs, a cofactor of HIV infection, as well as self-reported HIV infection and perceptions of risk for infection.