Finite mixture models have offered a very important tool for exploring complex data structures in many scientific areas, such as economics, epidemiology and finance. Semiparametric models, which were introduced into traditional finite the past decade, brought forth exciting developments their methodologies, theories, applications. In this article, we not only provide selective overview of newly-developed semiparametric but also discuss estimation theoretical properties if applicable, some...

In this article, we propose a novel nonparametric statistical learning tool based on modal regression, which can complement the standard mean and quantile regression has broad applicability in various fields. We first local polynomial focuses most likely conditional value (conditional mode) of dependent variable Y given covariates x, several superiorities over or quantiles, such as resistance to outliers some forms measurement error having shorter prediction intervals when data are skewed....

Abstract In this paper, we propose a new effective estimator for class of semiparametric mixture models where one component has known distribution with possibly unknown parameters while the other density and mixing proportion are unknown. Such have been often used in multiple hypothesis testing sequential clustering algorithm. The proposed is based on minimum profile Hellinger distance (MPHD), its theoretical properties investigated. addition, use simulation studies to illustrate finite...

Variability in ADG of feedlot cattle can affect profits, thus making overall returns more unstable. Hence, knowledge the factors that contribute to heterogeneity variances animal performance help managers evaluate risks and minimize profit volatility when managerial economic decisions commercial feedlots. The objectives present study were heteroskedasticity, defined as variances, cohorts cattle, identify demographic at arrival potential sources variance heterogeneity, accounting for cohort-...

In this article, we propose an efficient and robust estimation for the semiparametric mixture model that is a of unknown location-shifted symmetric distributions. Our derived by minimizing profile Hellinger distance (MPHD) between nonparametric density estimate. We simple algorithm to find proposed MPHD estimation. Monte Carlo simulation study conducted examine finite sample performance procedure compare it with other existing methods. Based on our empirical studies, newly works very...

Mixtures of factor analyzers (MFAs) have been popularly used to cluster the high-dimensional data. However, traditional estimation method is based on normality assumptions random terms and thus sensitive outliers. In this article, we introduce a robust procedure MFAs using trimmed likelihood estimator. We use simulation study real data application demonstrate robustness compare it with normality-based maximum estimate.

In this article, we propose a new nonparametric data analysis tool, which call modal regression, to investigate the relationship among interested variables based on estimating mode of conditional density response variable Y given predictors X. The regression is distinguished from conventional in that, instead average or median, it uses "most likely" values measures center. Better prediction performance and robustness are two important characteristics compared traditional mean median...

Abstract. To help the lender to make reasonable prediction judgments on in advance, get most appropriate way of lending amount, as well facing risk a response program and ability cope with need predictions advance before for personal credit risk. This paper aims predict default based random forest, logistic regression decision tree algorithms, by comparing analyzing advantages disadvantages these this finally chooses forest algorithm. concludes that predicting default, three characteristics...

In this article, we propose two classes of semiparametric mixture regression models with single-index for model based clustering. Unlike many semiparametric/nonparametric that can only be applied to low dimensional predictors, the new easily incorporate high predictors into nonparametric components. The proposed are very general, and recently indeed special cases models. Backfitting estimates corresponding modified EM algorithms achieve optimal convergence rates both parametric parts. We...

Semiparametric mixture of regression (SMR) models provide a popular and flexible framework for modeling heterogeneous data that violates some the parametric assumptions assumed in traditional finite regressions models. The majority applications SMR assume normality their error terms. As is well known, Gaussian distribution sensitive to outliers or heavy-tailed distribution. In this article, we propose more robust approach by as t distributions. By combining EM algorithm kernel density...