Principal Component Logistic Regression Research Articles

Epilepsy EEG signals classification based on sparse principal component logistic regression model

Epilepsy is a chronic brain disease caused by excessive discharge of brain neurons. Long-term recurrent seizures bring a lot of trouble to patients and their families. Prediction of different stages of epilepsy is of great significance. We extract pearson correlation coefficients (PCC) between channels in different frequency bands as features of EEG signals for epilepsy stages prediction. However, the features are of large feature dimension and serious multi-collinearity. To eliminate these adverse influence, the combination of traditional dimension reduction method principal component analysis (PCA) and logistic regression method with regularization term is proposed to avoid over-fitting and achieve the feature sparsity. The experiments are conducted on the widely used CHB-MIT dataset using different regularization terms L 1 and L 2 , respectively. The proposed method identifies various stages of epilepsy quickly and efficiently, and it presents the best average accuracy of 94.86%, average precision of 96.71%, average recall of 93.48%, average kappa value of 0.90 and average Matthews correlation coefficient (MCC) value of 0.90 for all patients.

Effects of Air Pollution Exposure during Preconception and Pregnancy on Gestational Diabetes Mellitus.

This study aimed to investigate the association between air pollution and gestational diabetes mellitus (GDM) in small- and medium-sized cities, identify sensitive periods and major pollutants, and explore the effects of air pollution on different populations. A total of 9820 women who delivered in Handan Maternal and Child Health Hospital in the Hebei Province from February 2018 to July 2020 were included in the study. Logistic regression and principal component logistic regression models were used to assess the effects of air pollution exposure during preconception and pregnancy on GDM risk and the differences in the effects across populations. The results suggested that each 20 μg/m3 increase in PM2.5 and PM10 exposure during preconception and pregnancy significantly increased the risk of GDM, and a 10 μg/m3 increase in NO2 exposure during pregnancy was also associated with the risk of GDM. In a subgroup analysis, pregnant women aged 30-35 years, nulliparous women, and those with less than a bachelor's education were the most sensitive groups. This study provides evidence for an association between air pollution and the prevalence of GDM, with PM2.5, PM10, and NO2 as risk factors for GDM.

Classification and prediction of the chemical composition of glass based on grey and principal component Logistic regression

As a momentous witness of the ancient Silk Road trade, glass’s variety identification and measurement or even prediction of its chemical constituent content are significantly important for people to have a systematic understanding of it. However, glass weathering tends to occur because of the ambient conditions of the place where it is buried, which may bring some difficulties to the classification and content prediction. Based on the data provided in problem C of the National Undergraduate Mathematical Modeling Contest in 2022, the GM (1,1) grey prediction model has first constructed so that the content ranges of various pivotal components are achieved. Then several glass samples whose category is unknown could be classified based on a principal component logistic regression model of the chemical composition’s content.

Relationship of long-term air pollution exposure with asthma and rhinitis in Italy: an innovative multipollutant approach

Backgroundair pollution is a complex mixture; novel multipollutant approaches could help understanding the health effects of multiple concomitant exposures to air pollutants. Aimto assess the relationship of long-term air pollution exposure with the prevalence of respiratory/allergic symptoms and diseases in an Italian multicenter study using single and multipollutant approaches. Methods14420 adults living in 6 Italian cities (Ancona, Pavia, Pisa, Sassari, Turin, Verona) were investigated in 2005–2011 within 11 different study cohorts. Questionnaire information about risk factors and health outcomes was collected. Machine learning derived mean annual concentrations of PM10, PM2.5, NO2 and mean summer concentrations of O3 (μg/m3) at residential level (1-km resolution) were used for the period 2013–2015. The associations between the four pollutants and respiratory/allergic symptoms/diseases were assessed using two approaches: a) logistic regression models (single-pollutant models), b) principal component logistic regression models (multipollutant models). All the models were adjusted for age, sex, education level, smoking habits, season of interview, climatic index and included a random intercept for cohorts. Resultsthe three-year average (± standard deviation) pollutants concentrations at residential level were: 20.3 ± 6.8 μg/m3 for PM2.5, 29.2 ± 7.0 μg/m3 for PM10, 28.0 ± 11.2 μg/m3 for NO2, and 70.9 ± 4.3 μg/m3 for summer O3. Through the multipollutant models the following associations emerged: PM10 and PM2.5 were related to 14–25% increased odds of rhinitis, 23–34% of asthma and 30–33% of night awakening; NO2 was related to 6–9% increased odds of rhinitis, 7–8% of asthma and 12% of night awakening; O3 was associated with 37% increased odds of asthma attacks. Overall, the Odds Ratios estimated through the multipollutant models were attenuated when compared to those of the single-pollutant models. Conclusionsthis study enabled to obtain new information about the health effects of air pollution on respiratory/allergic outcomes in adults, applying innovative methods for exposure assessment and multipollutant analyses.

Multinomial Principal Component Logistic Regression on Shape Data

Multinomial Principal Component Logistic Regression on Shape Data

Utilization of UV-visible and FTIR spectroscopy coupled with chemometrics for differentiation of Indonesian tea: an exploratory study

Ultraviolet (UV)-visible and Fourier-transformed infrared (FTIR) spectroscopy are two of the most popular and readily available laboratory instruments. Fingerprinting analysis of the UV-visible and FTIR spectra has been applied for food classification and authentication studies. In this study, the UV-visible and FTIR spectra of brewed tea, and their data fusion data sets, were used to build models for the classification of tea based on tea types and origins. The study included black and green tea samples from several provinces in Sumatra and Java Islands (Indonesia). Chemometric models of principal component analysis (PCA), k-nearest neighbor (kNN), and logistic regression were developed for classification purposes. All PCA models were able to well-separate the tea groups. kNN and logistic regression models based on UV-visible spectra successfully classified green and black tea with >0.8 classification accuracy. The kNN model of FTIR spectra had good accuracy (0.903) for classifying tea based on its origin. ReliefF algorithm was employed to select the best features among the data fusion data sets of UV-visible and FTIR spectra. The data fusion data sets of UV-visible and FTIR spectra demonstrated good separation of tea types and origins with a high area under the ROC curve (>0.8) and moderate accuracy (0.548). Therefore, UV-visible and FTIR spectroscopy may provide complementary information for tea classification based on tea types and origins.

Modeling go-around occurrence using principal component logistic regression

A go-around is an aborted approach of an aircraft. We model go-around occurrence using Principal Component Logistic Regression (PCLR). This entails go-around detection, feature engineering, and model estimation. As a case study, we consider John F. Kennedy (JFK) International Airport arrivals, and model go-around occurrence based on information available when the subject flight is five nautical miles from its landing runway threshold. The PCLR model is based on Principal Component Analysis (PCA) for analyzing data that suffer from multi-collinearity. The model provides a representation of the empirical relationship between go-around occurrence and Principal Components (PCs) covariates, which encompass flight approach features, aircraft characteristics, flight lead-trail spacing, surface operation, go-around clustering effect, airport and weather conditions. We use factor loading analysis to reveal the relationship between variables and the PCs they formed. Coefficient estimates of PCs are also transformed back to the scale of the original variables by matrix operations. A counterfactual analysis is employed to assess the importance of different features, and reveals that the stability of an approach, flight lead-trail spacing, departure traffic, and ceiling are the most salient factors affecting go-around occurrence.

Trends of HIV/Syphilis/HSV-2 seropositive rate and factors associated with HSV-2 infection in men who have sex with men in Shenzhen, China: A retrospective study.

ObjectivesTo analyze the trends of HIV/syphilis/HSV-2 seropositive rate and explore the related factors with HSV-2 infection to provide the basis for adjusting STD intervention strategies and formulating prevention and control measures among MSM in Shenzhen.MethodsTime-location sampling was conducted among MSM in Shenzhen in 2012, 2014, 2016, and 2018. Data on demographics, sexual behaviors and the laboratory test results of HIV, syphilis, HSV-2 were collected. The χ2 trend test was used to analyze the trends of HIV/syphilis/HSV-2 seropositive rate. The binary logistic regression model was used to explore the factors associated with HSV-2 infection.ResultsThe seropositive rate of HIV fell significantly from 15.9% in 2012 to 8.7% in 2018 (Ptrend = 0.003), syphilis seropositive rate was significantly decreased from 20.4% in 2012 to 14.8% in 2018 (Ptrend = 0.025), HSV-2 seropositive rate had no significant change (16.7% in 2012 to 14.0% in 2018; Ptrend = 0.617). In principal component logistic regression analysis showed that FAC1_1 (X1 = Ever had sex with female, X2 = Gender of first sexual partner, X3 = Marital status, X4 = Age group), FAC2_1 (X5 = Education, X6 = Monthly income (RMB), X7 = Frequency of condom use in anal sex with men in the past 6 months), and FAC4_1 (X9 = History of STDs) were significantly associated with HSV-2 infection.ConclusionsThe seropositive rates of HIV and syphilis have dropped significantly but are still high. HSV-2 seropositive rate had no significant change and maintained a high level. It is necessary to continue strengthening HIV and syphilis interventions among MSM in Shenzhen. HSV-2 detection and intervention are urgently required for MSM, which might be another effective biological strategy further to control the HIV epidemic among MSM in Shenzhen.

Risk evaluation of mine-water inrush based on principal component logistic regression analysis and an improved analytic hierarchy process

A method of evaluating the risk of floor water inrush using principal component logistic regression analysis (PCLRA) and an improved analytic hierarchy process (IAHP) is presented. The approach was validated by a case study at the Yangcheng coal mine in Shandong Province, China. First, the risk assessment index of floor water inrush was determined based on geological and hydrogeological conditions of the study area. Then, a comprehensive evaluation model (CEM), combining PCLRA with IAHP, was constructed to determine the comprehensive weight of each water-inrush evaluation index. Finally, water-inrush risk zoning was determined with GIS. The results show that the CEM, compared with the water-inrush coefficient method (WICM) traditionally and commonly used in China, has higher fitting accuracy and more detailed division of water-inrush risk areas. This method not only follows the observations in reality, but also fully considers the relative importance of water-inrush evaluation indices. The results can provide a theoretical basis for the safe mining of coal seams above confined groundwater.

Application of principal component analysis and logistic regression model in lupus nephritis patients with clinical hypothyroidism

BackgroundPrevious studies indicate that the prevalence of hypothyroidism is much higher in patients with lupus nephritis (LN) than in the general population, and is associated with LN’s activity. Principal component analysis (PCA) and logistic regression can help determine relevant risk factors and identify LN patients at high risk of hypothyroidism; as such, these tools may prove useful in managing this disease.MethodsWe carried out a cross-sectional study of 143 LN patients diagnosed by renal biopsy, all of whom had been admitted to Xiangya Hospital of Central South University in Changsha, China, between June 2012 and December 2016. The PCA–logistic regression model was used to determine the influential principal components for LN patients who have hypothyroidism.ResultsOur PCA–logistic regression analysis results demonstrated that serum creatinine, blood urea nitrogen, blood uric acid, total protein, albumin, and anti-ribonucleoprotein antibody were important clinical variables for LN patients with hypothyroidism. The area under the curve of this model was 0.855.ConclusionThe PCA–logistic regression model performed well in identifying important risk factors for certain clinical outcomes, and promoting clinical research on other diseases will be beneficial. Using this model, clinicians can identify at-risk subjects and either implement preventative strategies or manage current treatments.

Neighbourhood-level deprivation indices and postpartum women\u2019s health: results from the Community Child Health Network (CCHN) multi-site study

BackgroundArea-level socioeconomic characteristics have been shown to be related to health status and mortality however, little is known about the association between residential community characteristics in relation to postpartum women’s health.MethodsData from the longitudinal, multi-site Community Child Health Network (CCHN) study were used. Postpartum women (n = 2510), aged 18–40 were recruited from 2008 to 2012 within a month of delivery. Socioeconomic data was used to create deprivation indices. Census data were analysed using principal components analysis (PCA) and logistic regression to assess the association between deprivation indices (DIs) and various health indicators.ResultsPCA resulted in two unique DIs that accounted for 67.5% of the total variance of the combined all-site area deprivation. The first DI was comprised of variables representing a high percentage of Hispanic or Latina, foreign-born individuals, dense households (more than one person per room of residence), with less than a high-school education, and who spent more than 30% of their income on housing costs. The second DI was comprised of a high percentage of African-Americans, single mothers, and high levels of unemployment. In a multivariate logistic regression model, using the quartiles of each DI, women who reside in the geographic area of Q4-Q2 of the second DI, were almost twice as likely to have more than three adverse health conditions compared to those who resided in the least deprived areas. (Q2vs.Q1:OR = 2.09,P = 0.001,Q3vs.Q1:OR = 1.89,P = 0.006,Q4vs.Q1:OR = 1.95,P = 0.004 respectively).ConclusionsOur results support the utility of examining deprivation indices as predictors of maternal postpartum health.

Factors Affecting Successful Agricultural Loan Applications: The Case of a South African Credit Provider

The purpose of the paper is to determine the influence of different factors used by a formal credit institution to evaluate loan applications in the agricultural sector. The research attempts to capture the actual factors considered by credit institutions rather than the traditional factors found in literature. Loan applications from 128 farmers, predominantly commercial farmers, were obtained from a credit institution with branches situated in various provinces of South Africa. Data consisted of loan application information which is broader than the financial information normally obtained in credit research, and the final decision of the credit provider. Principal component logistic regression was used to investigate the likeliness with which loan application variables influence the outcome of the loan application. Results indicate that loan applications that are more likely to be successful are older more experienced farmers, who can provide sufficient collateral, have more years of business with the credit provider, have an acceptable credit history, request smaller loan amounts, have lower interest expense ratio, higher production cost ratios, and have diversification strategies. This paper contributes to knowledge on information used by financial credit providers (institutions) in classifying agricultural loan applications as successful as guided by actual factors used in credit decision making by the credit provider.

Identification of outlying and influential data with principal components regression estimation in binary logistic regression

In this study, we settle on the issue that when multicollinearity and unusual observations arise simultaneously and we straightforwardly extend leverages, Pearson residuals, delta beta and delta chi-square statistics using the principal components logistic regression (PCLR) estimator where the extensions typically take the advantage of the computation of PCLR estimator by one-step approximation. We then applied two simulation studies and a numerical example to illustrate the behavior of statistics for the PCLR estimator versus the traditional ML estimator.

A first-order approximated jackknifed ridge estimator in binary logistic regression

The purpose of this paper is to solve the problem of multicollinearity that affects the estimation of logistic regression model by introducing first-order approximated jackknifed ridge logistic estimator which is more efficient than the first-order approximated maximum likelihood estimator and has smaller variance than the first-order approximated jackknife ridge logistic estimator. Comparisons of the first-order approximated jackknifed ridge logistic estimator to the first-order approximated maximum likelihood, first-order approximated ridge, first-order approximated r-k class and principal components logistic regression estimators according to the bias, covariance and mean square error criteria are done. Three different estimators for the ridge parameter are also proposed. A real data set is used to see the performance of the first-order approximated jackknifed ridge logistic estimator over the first-order approximated maximum likelihood, first-order approximated ridge logistic, first-order approximated r-k class and first-order approximated principal components logistic regression estimators. Finally, two simulation studies are conducted in order to show the performance of the first-order approximated jackknife ridge logistic estimator.

Assessment of water inrush risk using the principal component logistic regression model in the Pandao coal mine, China

In order to improve the accuracy of floor water inrush assessment, the risk prediction model of floor water inrush was established by combining the principal component logistic regression analysis (PCLRA) and GIS spatial geographic analysis. In this paper, the geological data of Pandao coal mine was taken as the engineering background. First of all, main controlling factors of floor water inrush were determined and quantified. Next, PCLRA was used to determine the weight of each factor and establish the mathematical model for predicting the floor water inrush. And then, GIS’s spatial analysis and data processing function was used to draw related single factor thematic maps. Related thematic maps were weighted superposed to draw a floor water inrush zoning map based on PCLRA mathematical model. The study areas were divided into five levels by Jenks optimization method and vulnerability index initial model. And the corresponding threshold range was determined. The results show that (1) the high sensitivity factors in floor failure depth were added to evaluate the water inrush, and the fault fractal dimension was used to replace the fault structure related factors, and the main controlling factors of floor water inrush are more comprehensive; (2) the fitting degree of PCLRA model is high and the test accuracy is 83.3%; (3) the prediction results were well fitted to the actual position of water inrush (three water inrush points are located in the dangerous area, and two water inrush points are located in the relatively dangerous area).

Study on learning effect prediction models based on principal component analysis in MOOCs

In recent years, the rise and development of massive open online courses (MOOCs) have promoted the boom of online education and also promoted the research on learning analysis and mining based on big data of education. However, while offering a large number of high quality courses, there is also a phenomenon that the overall learning effect is not ideal. How to make effective use of MOOCs for teaching activities poses urgent practical requirements for educators and researchers. MOOCs store massive learners’ learning behavior data, and mining these data is of great significance to learners’ learning effects. Due to the deviation caused by the correlation between many behavior indexes, the paper analyze the nine measurable performance indexes by principal component analysis (PCA), and get three principal components factors in order to reduce the computational dimension and increase the comprehensibility, and then obtain a model of learning effect prediction, namely the PCA prediction model (PCA_PM). Logistic regression algorithm has a good fitting effect on behavioral indexes. Therefore, the paper use the logistic regression method to construct another model of learning effect prediction, namely the principal component logistic regression prediction model (PCL_PM), and predictive effects of two models were compared and analyzed. The results show that it is an effective way to help learners to improve the passing rate of the course by establishing the predictive models. The prediction accuracy of both PCA_PM and PCL_PM is high, but overall, the prediction effect of PCA_PM is better.

Development of Heavy Rain Damage Prediction Function Using Principal Component Analysis and Logistic Regression Model

Development of Heavy Rain Damage Prediction Function Using Principal Component Analysis and Logistic Regression Model

Multivariate geostatistical modeling of the spatial sediment distribution in a large scale drainage basin, Upper Rhone, Switzerland

There is a notable discrepancy between detailed sediment budget studies in small headwater catchments (<102km2) focusing on the identification of sedimentary landforms in the field (e.g. talus cones, moraine deposits, fans) and large scale studies (>103km2) in higher order catchments applying modeling and/or remote sensing based approaches for major sediment storage delineation.To bridge the gap between these scales, we compiled an inventory of sediment and bedrock coverage from field mapping, remote sensing analysis and published data for five key sites in the Upper Rhone Basin (Val d'Illiez, Val de la Liène, Turtmanntal, Lötschental, Goms; 360.3km2, equivalent to 6.7% of the Upper Rhone Basin). This inventory was used as training and testing data for the classification of sediment and bedrock cover. From a digital elevation model (2×2m ground resolution) and Landsat imagery we derived 22 parameters characterizing local morphometry, topography and position, contributing area, and climatic and biotic factors on different spatial scales, which were used as inputs for different statistical models (logistic regression, principal component logistic regression, generalized additive model). Best prediction results with an excellent performance (mean AUROC: 0.8721±0.0012) and both a high spatial and non-spatial transferability were achieved applying a generalized additive model. Since the model has a high thematic consistency, the independent input variables chosen based on their geomorphic relevance are suitable to model the spatial distribution of sediment.Our high-resolution classification shows that 53.5±21.7% of the Upper Rhone Basin are covered with sediment. These are by no means evenly distributed: small headwaters (<5km2) feature a very strong variability in sediment coverage, with watersheds drowning in sediments juxtaposed to watersheds devoid of sediment cover. In contrast, larger watersheds predominantly show a bimodal distribution, with highest densities for bedrock (30–40%) being consistently lower than for sediment cover (60–65%). Earlier studies quantifying sedimentary cover and volume focus on the broad glacially overdeepened Rhone Valley that accounts for c. 9% of our study area. While our data support its importance, we conservatively estimate that the remaining 90% of sediment cover, mainly located outside trunk valleys, account for a volume of 2.6–13km3, i.e. 2–16% of the estimated sediment volume stored in the Rhone Valley between Brig and Lake Geneva. Furthermore, our data reveal increased relative sediment cover in areas deglaciated since the Little Ice Age, as compared to headwater regions without this recent glacial imprint. We therefore conclude that sediment storage in low-order valleys, often neglected in large scale studies, constitutes a significant component of large scale sediment budgets that needs to be better included into future analysis.

Functional Data Analysis Applied to Modeling of Severe Acute Mucositis and Dysphagia Resulting From Head and Neck Radiation Therapy

PurposeCurrent normal tissue complication probability modeling using logistic regression suffers from bias and high uncertainty in the presence of highly correlated radiation therapy (RT) dose data. This hinders robust estimates of dose-response associations and, hence, optimal normal tissue—sparing strategies from being elucidated. Using functional data analysis (FDA) to reduce the dimensionality of the dose data could overcome this limitation.Methods and MaterialsFDA was applied to modeling of severe acute mucositis and dysphagia resulting from head and neck RT. Functional partial least squares regression (FPLS) and functional principal component analysis were used for dimensionality reduction of the dose-volume histogram data. The reduced dose data were input into functional logistic regression models (functional partial least squares—logistic regression [FPLS-LR] and functional principal component—logistic regression [FPC-LR]) along with clinical data. This approach was compared with penalized logistic regression (PLR) in terms of predictive performance and the significance of treatment covariate—response associations, assessed using bootstrapping.ResultsThe area under the receiver operating characteristic curve for the PLR, FPC-LR, and FPLS-LR models was 0.65, 0.69, and 0.67, respectively, for mucositis (internal validation) and 0.81, 0.83, and 0.83, respectively, for dysphagia (external validation). The calibration slopes/intercepts for the PLR, FPC-LR, and FPLS-LR models were 1.6/−0.67, 0.45/0.47, and 0.40/0.49, respectively, for mucositis (internal validation) and 2.5/−0.96, 0.79/−0.04, and 0.79/0.00, respectively, for dysphagia (external validation). The bootstrapped odds ratios indicated significant associations between RT dose and severe toxicity in the mucositis and dysphagia FDA models. Cisplatin was significantly associated with severe dysphagia in the FDA models. None of the covariates was significantly associated with severe toxicity in the PLR models. Dose levels greater than approximately 1.0 Gy/fraction were most strongly associated with severe acute mucositis and dysphagia in the FDA models.ConclusionsFPLS and functional principal component analysis marginally improved predictive performance compared with PLR and provided robust dose-response associations. FDA is recommended for use in normal tissue complication probability modeling.

Cytokine Profiles during Invasive Nontyphoidal Salmonella Disease Predict Outcome in African Children.

Nontyphoidal Salmonella is a leading cause of sepsis in African children. Cytokine responses are central to the pathophysiology of sepsis and predict sepsis outcome in other settings. In this study, we investigated cytokine responses to invasive nontyphoidal Salmonella (iNTS) disease in Malawian children. We determined serum concentrations of 48 cytokines with multiplexed immunoassays in Malawian children during acute iNTS disease (n = 111) and in convalescence (n = 77). Principal component analysis and logistic regression were used to identify cytokine signatures of acute iNTS disease. We further investigated whether these responses are altered by HIV coinfection or severe malnutrition and whether cytokine responses predict inpatient mortality. Cytokine changes in acute iNTS disease were associated with two distinct cytokine signatures. The first is characterized by increased concentrations of mediators known to be associated with macrophage function, and the second is characterized by raised pro- and anti-inflammatory cytokines typical of responses reported in sepsis secondary to diverse pathogens. These cytokine responses were largely unaltered by either severe malnutrition or HIV coinfection. Children with fatal disease had a distinctive cytokine profile, characterized by raised mediators known to be associated with neutrophil function. In conclusion, cytokine responses to acute iNTS infection in Malawian children are reflective of both the cytokine storm typical of sepsis secondary to diverse pathogens and the intramacrophage replicative niche of NTS. The cytokine profile predictive of fatal disease supports a key role of neutrophils in the pathogenesis of NTS sepsis.