Semiparametric Methods Research Articles

Semiparametric methods for incomplete longitudinal count data with an application to health and retirement study

In this paper, we propose and explore a novel semiparametric approach to analyzing longitudinal count data. We address the issue of missingness in longitudinal data and propose a weighted generalized estimations equations approach to fitting marginal mean response models for count responses with dropouts. Also, we investigate a spline regression approach to approximating the curvilinear relationship between the mean response and covariates. The asymptotic properties of the proposed estimators are studied in some detail. The empirical properties of the estimators are investigated using Monte Carlo simulations. An application is also provided using actual survey data obtained from the Health and Retirement Study (HRS).

Suggested methods for prediction using semiparametric regression function

Ferritin is a key organizer of protected deregulation, particularly below risky hyperferritinemia, by straight immune-suppressive and pro-inflammatory things. We conclude that there is a significant association between levels of ferritin and the harshness of COVID-19. In this paper, we introduce a semi-parametric method for prediction by making a combination of NN and regression models. So, two methodologies are adopted, Neural Network (NN) and regression model in designing the model;the data was collected from a nursing home hospital for period 11/7/2021-23/7/2021, the sample size is 100 covid positive patients with 12 females & 38 males out of 50, while 26 female & 24 male are non-COVID out of 50. The input variables of the NN model are identified as the ferritin and a gender variable. The higher results precision is attained by the multilayer perceptron (MLP) networks when we applied the explanatory variables as the inputs with one hidden layer, which covers 3 neurons, as the planned many hidden layers are with one output of the fitting NN model which is used in stages of training and validation beside the actual data. We used a portion of the actual data to verify the behavior of the developed models, we find out that only one observation is a false predictive value. This means that the estimation model has significant parameters to forecast the type of Covid cases (Covid or no Covid). © 2021, Semnan University, Center of Excellence in Nonlinear Analysis and Applications. All rights reserved.

Bayesian Clustering of Player Styles for Multiplayer Games

With game play data, empirical approaches to clustering are typically based solely on game outcomes, e.g. kills, deaths, and score for each player. In this paper, we investigate a method for clustering players based on how a player’s choices relate to outcomes, or equivalently the latent player styles exhibited by players. Our approach is based on a Bayesian semi-parametric clustering method which has several advantages: the number of clusters do not need to be specified a priori; the technique can work with a very compact representation of each match (e.g. consisting primarily of indicator variables for player choices); a player can belong to multiple clusters and hence can have a hybrid style; and the resulting clusterings often have a straight-forward interpretation. To demonstrate the approach, we apply our method to multiplayer match logs from Battlefield 3 consisting of over 1200 players and 500,000 matches.

Modelling extreme risks for carbon emission allowances — Evidence from European and Chinese carbon markets

Modelling the dynamics of carbon emission allowance prices has been of keen interest to academicians from around the world. The distribution of tails is different from that of centres for carbon allowance prices, invalidating regular methods and conventions applied in quantitative finance. Value-at-Risk (VaR) and Expected Shortfall (ES) are traditionally useful tools to measure the extreme risk of carbon markets. On this basis, this study uses the semiparametric method to estimate the time-varying VaR and ES of European Union Allowance (EUA) and Chinese carbon emission allowance (CEA) in downside and upside events. Our results show that the GAS model outperforms the rolling estimate in describing the time-varying downside and upside VaR and ES. We find two breaks in VaR and ES for EUA whilst no break for CEA. We also find the evidence of asymmetry in CEA and reversed asymmetry in EUA, but it differs across periods and between VaR and ES. In addition, the results reveal that the downside CEA VaR and ES are significantly greater than those of EUA. Meanwhile, the relationship between upside EUA and CEA series varies across periods as well as between VaR and ES. These findings have important implications for entrepreneurs, investors, risk managers, and policy makers.

An approximate quasi-likelihood approach for error-prone failure time outcomes and exposures.

Measurement error arises commonly in clinical research settings that rely on data from electronic health records or large observational cohorts. In particular, self-reported outcomes are typical in cohort studies for chronic diseases such as diabetes in order to avoid the burden of expensive diagnostic tests. Dietary intake, which is also commonly collected by self-report and subject to measurement error, is a major factor linked to diabetes and other chronic diseases. These errors can bias exposure-disease associations that ultimately can mislead clinical decision-making. We have extended an existing semiparametric likelihood-based method for handling error-prone, discrete failure time outcomes to also address covariate error. We conduct an extensive numerical study to compare the proposed method to the naive approach that ignores measurement error in terms of bias and efficiency in the estimation of the regression parameter of interest. In all settings considered, the proposed method showed minimal bias and maintained coverage probability, thus outperforming the naive analysis which showed extreme bias and low coverage. This method is applied to data from the Women's Health Initiative to assess the association between energy and protein intake and the risk of incident diabetes mellitus. Our results show that correcting for errors in both the self-reported outcome and dietary exposures leads to considerably different hazard ratio estimates than those from analyses that ignore measurement error, which demonstrates the importance of correcting for both outcome and covariate error.

Configurations of business founder resources, strategy, and environment determining new venture performance

ABSTRACT We advance a configurational perspective that assumes a new venture’s performance, operationalized as profitability in this study, is determined by interdependencies among its founder(s)’s resources (i.e., human, social, and financial capital), its strategy, and environmental munificence. Using Rule Ensembles, a semiparametric set-theoretic method, we conduct an abductive investigation using a sample of 853 new ventures. The Rule Ensembles technique identifies five multivariate interdependencies (rules) that enhance new venture profitability and two that reduce it. Indicators of business founder resources are involved in all of the seven model rules. Four configurational rules, including a strategy and/or environmental component, equally emerge.

Semiparametric estimation of the proportional rates model for recurrent events data with missing event category.

Proportional rates models are frequently used for the analysis of recurrent event data with multiple event categories. When some of the event categories are missing, a conventional approach is to either exclude the missing data for a complete-case analysis or employ a parametric model for the missing event type. It is well known that the complete-case analysis is inconsistent when the missingness depends on covariates, and the parametric approach may incur bias when the model is misspecified. In this paper, we aim to provide a more robust approach using a rate proportion method for the imputation of missing event types. We show that the log-odds of the event type can be written as a semiparametric generalized linear model, facilitating a theoretically justified estimation framework. Comprehensive simulation studies were conducted demonstrating the improved performance of the semiparametric method over parametric procedures. Multiple types of Pseudomonas aeruginosa infections of young cystic fibrosis patients were analyzed to demonstrate the feasibility of our proposed approach.

Estimating mean potential outcome under adaptive treatment length strategies in continuous time.

An adaptive treatment length strategy is a sequential stage-wise treatment strategy where a subject's treatment begins at baseline and one chooses to stop or continue treatment at each stage provided the subject has been continuously treated. The effects of treatment are assumed to be cumulative and, therefore, the effect of treatment length on clinical endpoint, measured at the end of the study, is of primary scientific interest. At the same time, adverse treatment-terminating events may occur during the course of treatment that require treatment be stopped immediately. Because the presence of a treatment-terminating event may be strongly associated with the study outcome, the treatment-terminating event is informative. In observational studies, decisions to stop or continue treatment depend on covariate history that confounds the relationship between treatment length on outcome. We propose a new risk-set weighted estimator of the mean potential outcome under the condition that time-dependent covariates update at a set of common landmarks. We show that our proposed estimator is asymptotically linear given mild assumptions and correctly specified working models. Specifically, we study the theoretical properties of our estimator when the nuisance parameters are modeled using either parametric or semiparametric methods. The finite sample performance and theoretical results of the proposed estimator are evaluated through simulation studies and demonstrated by application to the Enhanced Suppression of the Platelet Receptor IIb/IIIa with Integrilin Therapy (ESPRIT) infusion trial data.

Factor copula models for right-censored clustered survival data.

In this article we extend the factor copula model to deal with right-censored event time data grouped in clusters. The new methodology allows for clusters to have variable sizes ranging from small to large and intracluster dependence to be flexibly modeled by any parametric family of bivariate copulas, thus encompassing a wide range of dependence structures. Incorporation of covariates (possibly time dependent) in the margins is also supported. Three estimation procedures are proposed: both one- and two-stage parametric and a two-stage semiparametric method where marginal survival functions are estimated by using a Cox proportional hazards model. We prove that the estimators are consistent and asymptotically normally distributed, and assess their finite sample behavior with simulation studies. Furthermore, we illustrate the proposed methods on a data set containing the time to first insemination after calving in dairy cattle clustered in herds of different sizes.

Sample selection models with monotone control functions

The celebrated Heckman selection model yields a selection correction function (control function) proportional to the inverse Mills ratio, which is monotone. This paper studies a sample selection model that does not impose parametric distributional assumptions on the latent error terms, while maintaining the monotonicity of the control function. We show that a positive (negative) dependence condition on the latent error terms is sufficient for the monotonicity of the control function. The condition is equivalent to a restriction on the copula function of latent error terms. Using the monotonicity, we propose a tuning-parameter-free semiparametric estimation method and establish root n-consistency and asymptotic normality for the estimates of finite-dimensional parameters. A new test for selectivity is also developed in the presence of the shape restriction. Simulations and an empirical application are conducted to illustrate the usefulness of the proposed methods.

Weibull Regression and Stratified Cox Regression in Modelling Exclusive Breastfeeding Duration

Survival analysis has three approaches, parametric, non-parametric, and semi-parametric. The parametric model requires the distribution of survival time to be known. Weibull regression is one of the most popular forms of parametric approaches with advantages in flexibility and simplicity of hazard functions and survival functions. Stratified Cox regression is a semi-parametric approach method that is rightly used when the proportional hazard assumption is violated. The comparison between Weibull regression and stratified Cox regression was applied in the case of the duration of exclusive breastfeeding in infants aged 0-6 months in Indonesia. Before the model was formed, the data were tested for the Weibull distribution and the results were met properly. Whereas for testing proportional hazard assumptions the results are not fulfilled so that semi-parametric models can be used. Based on the results of the study, the Weibull regression is better than the stratified Cox regression.

Generic and specific recurrent neural network models: Applications for large and small scale biopharmaceutical upstream processes

The calculation of temporally varying upstream process outcomes is a challenging task. Over the last years, several parametric, semi-parametric as well as non-parametric approaches were developed to provide reliable estimates for key process parameters. We present generic and product-specific recurrent neural network (RNN) models for the computation and study of growth and metabolite-related upstream process parameters as well as their temporal evolution. Our approach can be used for the control and study of single product-specific large-scale manufacturing runs as well as generic small-scale evaluations for combined processes and products at development stage. The computational results for the product titer as well as various major upstream outcomes in addition to relevant process parameters show a high degree of accuracy when compared to experimental data and, accordingly, a reasonable predictive capability of the RNN models. The calculated values for the root-mean squared errors of prediction are significantly smaller than the experimental standard deviation for the considered process run ensembles, which highlights the broad applicability of our approach. As a specific benefit for platform processes, the generic RNN model is also used to simulate process outcomes for different temperatures in good agreement with experimental results. The high level of accuracy and the straightforward usage of the approach without sophisticated parameterization and recalibration procedures highlight the benefits of the RNN models, which can be regarded as promising alternatives to existing parametric and semi-parametric methods.

Abstract P116: Higher Body Mass Index Trajectories Are Associated With Lower Levels Of Physical Activity Measured By A Smartwatch

Introduction: The prevalence of obesity is rising. Most previous studies that examined the relationship between body mass index (BMI) and physical activity measured BMI at a single time-point, ignoring the time-varying nature of BMI. The relationship between BMI trajectories and habitual physical activity in community settings remains unclear. Objective: To assess the relationship between BMI trajectories and habitual physical activity measured by daily steps from a smartwatch, among participants enrolled in the electronic Framingham Heart Study (eFHS). We hypothesized that participants whose BMI trajectories increased over a 14-year period prior to the step assessment take fewer daily steps, compared to participants who maintained stable BMI trajectories during the same time period. Methods: We used a semiparametric group-based modelling method to identify BMI trajectory patterns. Participants who attended exams 1, 2, and 3 were included in building the trajectories. Daily steps were recorded from the smartwatch provided at exam 3 with “active days” defined as days with ≥ 5watch wear-hours. We excluded participants with <30 active days. The median follow-up period for step count was 357 days (IQR: 467 days). We used generalized linear models that accounted for correlation between daily steps in the same individuals to examine the longitudinal relationship between BMI trajectory groups and daily step counts, adjusting for relevant covariates. Results: We identified three trajectory groups for the 837 eFHS participants. Group 1 included 292 participants (mean age 54 years, 57% women) whose BMI was stable (slope: 0.005, p=0.75); Group 2 included 468 participants (mean age 53 years, 56% women) whose BMI increased slightly (slope: 0.123, p<9.2e-17); and Group 3 included 77 participants (mean age 50 years, 70% women) who had the largest increase of BMI (slope: 0.318, p=2.8e-22).Adjusting for age, sex, wear time and race/ethnicity, participants in group 3 (Δ1437 steps P< 0.0001) and Group 2 (Δ422 steps, P=0.04) took significantly fewer steps, compared to participants in Group 1 (Model 1). The effect sizes were slightly attenuated but remained significant after additionally adjusting for hypertension, type 2 diabetes, current smoking, and cardiovascular disease: Group 3 took 1258 fewer steps, P=0.0001; Group 2 took 406 fewer steps, P=0.04 (Model 2). We further adjusted for sleep apnea, education, and marital status in Model 3 and observed that on average Group 3 took 1120 fewer steps (P= 0.0007) and Group 2 took 382 fewer steps (P= 0.06), compared to Group 1. Conclusion: Participants whose BMI trajectory increased over time took significantly fewer steps compared to participants with more stable BMI trajectories. Our findings suggest that levels of physical activity may correlate with greater weight gain during adulthood.

On the effectiveness of case management for people with disabilities

Case managers provide individual and comprehensive support to employees who have become incapable of working. Using data from a large insurance company we find that overall, 43.9% of the people in our sample could be reintegrated. Controlling for personal characteristics, we analyze the effectiveness of case management by modelling the probability of reintegrating people being incapable of working into the labor market. Using parametric and semiparametric decomposition methods, we control for observational differences. We analyze how much of the difference in the reintegration rate between people who participate in case management and those who do not, is due to differences in characteristics and how much is due to case management itself. We find that the estimated probability of reintegration is 18.9% higher if people participate in case management. Moreover, our results show that no more than 15% are due to differences in characteristics and at least 85% can be attributed to case management itself.

New semiparametric regression method with applications in selection‐biased sampling and missing data problems

We propose a new method to estimate a regression function based on the semiparametric density ratio model, which can be viewed as a generalized linear model with a canonical link function and an unspecified baseline distribution function. Under this model, the distribution of the observed data retains the same structure in the presence of selection‐biased sampling or when the predictors are missing at random. In particular, in the latter case, the new method utilizes all the available information and does not need to specify the distribution of the predictors or the probability of observing the predictors. We establish large sample properties of the proposed regression estimators. Simulation studies demonstrate that the proposed estimators perform well in practical situations. Empirical data from the National Health and Nutrition Examination Survey are presented.

Semiparametric Bayesian Methods in Growth Curve Modeling for Nonnormal Data Analysis

Semiparametric Bayesian methods have been proposed in the literature for growth curve modeling to reduce the adverse effect of having nonnormal data. The normality assumption of measurement errors in traditional growth curve models was {replaced} by a random distribution with Dirichlet process mixture priors. However, both the random effects and measurement errors are equally likely to be nonnormal. Therefore, in this study, three types of robust distributional growth curve models are proposed from a semiparametric Bayesian perspective, in which random coefficients or measurement errors follow either normal distributions or unknown random distributions with Dirichlet process mixture priors. Based on a Monte Carlo simulation study, we evaluate the performance of the robust models and demonstrate that selecting an appropriate model for practical data analyses is very important, by comparing the three types of robust distributional models as well as the traditional growth curve models with the normality assumption. We also provide a straightforward strategy to select the appropriate model.

Outliers in Semi-Parametric Estimation of Treatment Effects

Outliers can be particularly hard to detect, creating bias and inconsistency in the semi-parametric estimates. In this paper, we use Monte Carlo simulations to demonstrate that semi-parametric methods, such as matching, are biased in the presence of outliers. Bad and good leverage point outliers are considered. Bias arises in the case of bad leverage points because they completely change the distribution of the metrics used to define counterfactuals; good leverage points, on the other hand, increase the chance of breaking the common support condition and distort the balance of the covariates, which may push practitioners to misspecify the propensity score or the distance measures. We provide some clues to identify and correct for the effects of outliers following a reweighting strategy in the spirit of the Stahel-Donoho (SD) multivariate estimator of scale and location, and the S-estimator of multivariate location (Smultiv). An application of this strategy to experimental data is also implemented.

Nonlinear autoregressive stochastic frontier model with dynamic technical inefficiency in panel data

In this paper, we focused on the new stochastic frontier model with the nonlinear autoregressive structure in panel data. In classic finance models, uncorrelated errors are considered, which are frequently not met in empirical situations. In our stochastic frontier model, the composite error consists of two components, the statistical error and technical inefficiency, whereas the technical inefficiency is assumed to be autocorrelated. A semiparametric method is suggested to estimate nonlinear autoregressive function by two steps procedure as parametric Taylor series expansion and nonparametric adjustment factor. For the model parameters estimation, the Expectation-Maximization approach is applied and the performance of the estimation is checked by Monte Carlo simulation. Finally, we investigate the applicability of new model using a real data set.

English

Family farms are poorly modernized in Burkina Faso despite their predominance in the country’s agriculture and their major contribution to national food production. Convincing evidence of the contribution of family farm modernization to food security is needed to support advocacy. This study used data from recent national longitudinal surveys and Cox semi-parametric regression methods to explore the effect of factors of modernization on the food security of farm households in Burkina Faso. The results showed that the training of agricultural workers, ownership of traction animals, and use of improved seeds reduced the risk of food-secure households falling into food insecurity by 22.8, 21.6, and 14.9%, respectively. These three factors significantly determine the stability of households’ food security, suggesting that the modernization of family farms could contribute to the prevention of food insecurity in Burkina Faso. A key strength of this study is that it was able to capitalize on the wealth of these data, which come from national surveys that are representative of farm households at the provincial level, longitudinal and prospective, making it possible to track the same households over time, at an annual frequency. Key words: Agricultural modernization, family farming, food security, Burkina Faso

Forecasting Unemployment Using Internet Search Data via PRISM

Big data generated from the Internet offer great potential for predictive analysis. Here we focus on using online users’ Internet search data to forecast unemployment initial claims weeks into the future, which provides timely insights into the direction of the economy. To this end, we present a novel method Penalized Regression with Inferred Seasonality Module (PRISM), which uses publicly available online search data from Google. PRISM is a semiparametric method, motivated by a general state-space formulation, and employs nonparametric seasonal decomposition and penalized regression. For forecasting unemployment initial claims, PRISM outperforms all previously available methods, including forecasting during the 2008–2009 financial crisis period and near-future forecasting during the COVID-19 pandemic period, when unemployment initial claims both rose rapidly. The timely and accurate unemployment forecasts by PRISM could aid government agencies and financial institutions to assess the economic trend and make well-informed decisions, especially in the face of economic turbulence.