Real Data Analysis Research Articles

토픽모델링에 기반한 R&D 투자 포트폴리오 분석 및 예측

Many countries are identifying crucial technologies for national security and economic growth amid global tech competition, crafting cross-departmental strategies to pursue them. We propose a platform for analyzing and predicting R&D investment portfolios based on government R&D project information and investment data to provide scientific evidence for strategic budget allocation. The proposed platform consists of the following four processes. First, topic modeling and project labeling are performed using project-related text data. Second, we propose a method for calculating the importance of each label based on the centered log ratio (CLR) transformation. Third, we predict the label importance for the next year using linear regression and Holt's linear trend method. Fourth, we propose a method for constructing investment portfolio in specific fields. Real data analysis demonstrates the explanability and predictability of the proposed platform by using research project data spanning 11 years from 2012 to 2022. We believe a novelty of the proposed method can contribute to the establishment of data-driven R&D investment portfolios.

Test and Measure for Partial Mean Dependence Based on Machine Learning Methods

It is of importance to investigate the significance of a subset of covariates W for the response Y given covariates Z in regression modeling. To this end, we propose a significance test for the partial mean independence problem based on machine learning methods and data splitting. The test statistic converges to the standard Chi-squared distribution under the null hypothesis while it converges to a normal distribution under the fixed alternative hypothesis. Power enhancement and algorithm stability are also discussed. If the null hypothesis is rejected, we propose a partial Generalized Measure of Correlation (pGMC) to measure the partial mean dependence of Y given W after controlling for the nonlinear effect of Z. We present the appealing theoretical properties of the pGMC and establish the asymptotic normality of its estimator with the optimal root-N convergence rate. Furthermore, the valid confidence interval for the pGMC is also derived. As an important special case when there are no conditional covariates Z, we introduce a new test of overall significance of covariates for the response in a model-free setting. Numerical studies and real data analysis are also conducted to compare with existing approaches and to demonstrate the validity and flexibility of our proposed procedures. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.

Tyranny-of-the-Minority Regression Adjustment in Randomized Experiments

Regression adjustment is widely used in the analysis of randomized experiments to improve the estimation efficiency of the treatment effect. This article reexamines a weighted regression adjustment method termed tyranny-of-the-minority (ToM), wherein units in the minority group are given greater weights. We demonstrate that ToM regression adjustment is more robust than Lin’s regression adjustment with treatment-covariate interactions, even though these two regression adjustment methods are asymptotically equivalent in completely randomized experiments. Moreover, ToM regression adjustment can be easily extended to stratified randomized experiments and completely randomized survey experiments. We obtain the design-based properties of the ToM regression-adjusted average treatment effect estimator under such designs. In particular, we show that the ToM regression-adjusted estimator improves the asymptotic estimation efficiency compared to the unadjusted estimator, even when the regression model is misspecified, and is optimal in the class of linearly adjusted estimators. We also study the asymptotic properties of various heteroscedasticity-robust standard errors and provide recommendations for practitioners. Simulation studies and real data analysis demonstrate ToM regression adjustment’s superiority over existing methods. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.

Robust estimate for count time series using GLARMA models: An application to environmental and epidemiological data

The Generalized Linear Autoregressive Moving Average (GLARMA) model has been used in epidemiological studies to evaluate the impact of air pollutants on health. Due to the nature of the data, a robust approach for the GLARMA model is proposed here based on the robustification of the quasi-likelihood function. Outlying observations are bounded separately by weight functions on covariates and the Huber loss function on the response variable. Some technical issues related to the robust approach are discussed and a Monte Carlo study revealed that the robust approach is more reliable than the classic one for contaminated data with additive outliers. The real data analysis investigates the impact of PM10 in the number of deaths by respiratory diseases in Vitória, Brazil.

Estimation in functional partially linear spatial autoregressive model

Functional regression has been a hot topic in statistical research. However, not much work has been done when response variables are cross-sectionally dependent variables and explanatory variables contain a real-valued scalar variable and a functional-valued random variable. In this paper, we consider a new functional partially linear spatial autoregressive model. Based on the functional principal components analysis and basis function approximation, we obtain the estimators of the unknown parameter and functions through the instrumental variables estimation method. The asymptotic normality and convergence rates of estimators are proved under some mild conditions. In addition, we illustrate the finite sample performance of the proposed estimation method through simulation study and a real data analysis.

Multi-Frequency Entropy for Quantifying Complex Dynamics and Its Application on EEG Data.

Multivariate entropy algorithms have proven effective in the complexity dynamic analysis of electroencephalography (EEG) signals, with researchers commonly configuring the variables as multi-channel time series. However, the complex quantification of brain dynamics from a multi-frequency perspective has not been extensively explored, despite existing evidence suggesting interactions among brain rhythms at different frequencies. In this study, we proposed a novel algorithm, termed multi-frequency entropy (mFreEn), enhancing the capabilities of existing multivariate entropy algorithms and facilitating the complexity study of interactions among brain rhythms of different frequency bands. Firstly, utilizing simulated data, we evaluated the mFreEn's sensitivity to various noise signals, frequencies, and amplitudes, investigated the effects of parameters such as the embedding dimension and data length, and analyzed its anti-noise performance. The results indicated that mFreEn demonstrated enhanced sensitivity and reduced parameter dependence compared to traditional multivariate entropy algorithms. Subsequently, the mFreEn algorithm was applied to the analysis of real EEG data. We found that mFreEn exhibited a good diagnostic performance in analyzing resting-state EEG data from various brain disorders. Furthermore, mFreEn showed a good classification performance for EEG activity induced by diverse task stimuli. Consequently, mFreEn provides another important perspective to quantify complex dynamics.

Exact interval estimation for three parameters subject to false positive misclassification

SummaryBinary data subject to one type of misclassification exist in various fields. It is collected in a double‐sampling scheme that includes a gold standard test and a fallible test. The main parameter of interest for this type of data is the positive probability of the gold standard test. Existing intervals are unreliable because the given nominal level is not achieved. In this paper, we construct an exact interval by inverting the E+M score tests and improve it by the general ‐function method. We find that the total length of the improved interval is shorter than the exact intervals that are also the improved intervals when we apply the ‐function to several existing approximate intervals, including the score and Bayesian intervals. Therefore, it is recommended for practice. We are also interested in two other parameters: —the difference between the two positive rates for the fallible and gold standard tests—and —the false positive rate for the fallible test. To the best of our knowledge, the research on these two parameters is limited. For , we find that any interval for can be converted to an interval for . So, the interval converted from the aforementioned recommended interval for is recommended for inferring . For , the improved interval by the ‐function method over the E+M score interval is derived. We use an example to illustrate how the intervals are computed and provide a real data analysis.

Sharma–Taneja–Mittal Entropy and Its Application of Obesity in Saudi Arabia

This paper presents several nonparametric estimators for the Sharma–Taneja–Mittal entropy measure of a continuous random variable with known support, utilizing spacing, a local linear model, and a kernel function. The properties of these estimators are discussed. Their performance was also examined through real data analysis and Monte Carlo simulations. In the Monte Carlo experiments, the proposed Sharma–Taneja–Mittal entropy estimators were employed to create a test of goodness-of-fit under the standard uniform distribution. The suggested test statistics demonstrate strong performance, as evidenced by a comparison of their power with that of other tests for uniformity. Finally, we examine a classification issue in the recognition of patterns to underscore the significance of these measures.

Wasserstein principal component analysis for circular measures

We consider the 2-Wasserstein space of probability measures supported on the unit-circle, and propose a framework for Principal Component Analysis (PCA) for data living in such a space. We build on a detailed investigation of the optimal transportation problem for measures on the unit-circle which might be of independent interest. In particular, building on previously obtained results, we derive an expression for optimal transport maps in (almost) closed form and propose an alternative definition of the tangent space at an absolutely continuous probability measure, together with fundamental characterizations of the associated exponential and logarithmic maps. PCA is performed by mapping data on the tangent space at the Wasserstein barycentre, which we approximate via an iterative scheme, and for which we establish a sufficient a posteriori condition to assess its convergence. Our methodology is illustrated on several simulated scenarios and a real data analysis of measurements of optical nerve thickness.

Natural disease course of chronic visceral acid sphingomyelinase deficiency in adults: A first step toward treatment criteria.

Acid sphingomyelinase deficiency (ASMD) is an ultra-rare lysosomal storage disease with a broad spectrum of manifestations ranging from severe neuropathic forms to attenuated, chronic visceral forms. Manifestations of the chronic visceral subtype are variable and encompass different degrees of hepatosplenomegaly, pulmonary disease and dyslipidemia. The aim of this study was to provide insights into the natural course of adult patients with the chronic visceral subtype. Based on these insights, we proposed tentative criteria for initiation and follow-up of enzyme replacement therapy (ERT). The data of 23 adult patients were collected in a prospective study. Clinical, genetic and demographic data, plasma measurements, abdominal imaging, pulmonary imaging, pulmonary function tests and quality of life questionnaires were collected. Stability of disease based on several clinical, biochemical and radiological markers (i.e., spleen volume, platelet levels, liver volume, alanine aminotransferase [ALT] levels, diffusion capacity of the lungs for carbon monoxide [DLCO] chitotriosidase activity and lysosphingomyelin [LSM]) was assessed. Cardiovascular risk was estimated based on sex, age, smoking, systolic blood pressure and lipid profile. Quality of life was evaluated with the 36-Item Short Form Health Survey and the Health Assessment Questionnaire. Median follow-up was 6.1 years (range 1.3-19.5 years). The most common manifestations were splenomegaly (100%), decreased high-density lipoprotein cholesterol (HDL-C) plasma levels (83%), (signs of) steatosis measured with transient elastography (82%), thrombocytopenia (64%), hepatomegaly (52%) and decreased diffusion capacity (45%). The majority of markers remained stable during follow-up. Twelve patients showed progression of disease: four for spleen volume, two for liver volume, three for DLCO, seven for chitotriosidase activity and three for LSM. One patient showed progression of disease based on four markers, although this patient did not report any problems at the last visit. Cardiovascular risk was estimated and was increased in half of the patients older than 40 years. Patient-reported quality of life did not differ from the general population, but differences in median 36-Item Short Form Health Survey (SF-36) scores of patients with severe pulmonary involvement and those of patients without pulmonary involvement were observed. Tentative criteria for initiation and effect of therapy were proposed. In conclusion, the chronic visceral subtype of ASMD showed a predominantly stable disease course in this cohort. We propose that ERT should be initiated on an individual basis and only in case of progression or symptomatic disease. Collection and analysis of real world data are necessary to refine start, stop and follow-up criteria in the future.

Minimax rates of convergence for sliced inverse regression with differential privacy

Sliced inverse regression (SIR) is a highly efficient paradigm used for the purpose of dimension reduction by replacing high-dimensional covariates with a limited number of linear combinations. This paper focuses on the implementation of the classical SIR approach integrated with a Gaussian differential privacy mechanism to estimate the central space while preserving privacy. We illustrate the tradeoff between statistical accuracy and privacy in sufficient dimension reduction problems under both the classical low- dimensional and modern high-dimensional settings. Additionally, we achieve the minimax rate of the proposed estimator with Gaussian differential privacy constraint and illustrate that this rate is also optimal for multiple index models with bounded dimension of the central space. Extensive numerical studies on synthetic data sets are conducted to assess the effectiveness of the proposed technique in finite sample scenarios, and a real data analysis is presented to showcase its practical application.

Interval-Censored Linear Quantile Regression

Censored quantile regression has emerged as a prominent alternative to classical Cox’s proportional hazards model or accelerated failure time model in both theoretical and applied statistics. While quantile regression has been extensively studied for right-censored survival data, methodologies for analyzing interval-censored data remain limited in the survival analysis literature. This article introduces a novel local weighting approach for estimating linear censored quantile regression, specifically tailored to handle diverse forms of interval-censored survival data. The estimation equation and the corresponding convex objective function for the regression parameter can be constructed as a weighted average of quantile loss contributions at two interval endpoints. The weighting components are nonparametrically estimated using local kernel smoothing or ensemble machine learning techniques. To estimate the nonparametric distribution mass for interval-censored data, a modified EM algorithm for nonparametric maximum likelihood estimation is employed by introducing subject-specific latent Poisson variables. The proposed method’s empirical performance is demonstrated through extensive simulation studies and real data analyses of two HIV/AIDS datasets. Supplementary materials for this article are available online.

Interaction screening in high-dimensional multi-response regression via projected distance correlation

Interaction screening for high-dimensional data is a challenging issue, especially for the strongly correlated predictors. A new two-stage interaction screening procedure based on the projected distance correlation is proposed when the predictors are highly correlated. To remove the confounding effect from the target variable that is induced by its correlated variables, we project the predictors and responses onto a conditional set. Our method can successfully identify important variables when the variables are highly correlated, and it can also identify variables that make a contribution to the response conditionally but not marginally. Moreover, our method is computationally efficient and simple, generally applicable without the requirement of the heredity assumption. Theoretical results show that the proposed method can yield the sure screening property. Simulation studies and real data analysis demonstrate the utility and validity of our method.

Grouped feature screening for ultrahigh-dimensional classification via Gini distance correlation

Gini distance correlation (GDC) was recently proposed to measure the dependence between a categorical variable, Y, and a numerical random vector, X. It mutually characterizes independence between X and Y. In this article, we utilize the GDC to establish a feature screening for ultrahigh-dimensional discriminant analysis where the response variable is categorical. It can be used for screening individual features as well as grouped features. The proposed procedure possesses several appealing properties. It is model-free. No model specification is needed. It holds the sure independence screening property and the ranking consistency property. The proposed screening method can also deal with the case that the response has divergent number of categories. We conduct several Monte Carlo simulation studies to examine the finite sample performance of the proposed screening procedure. Real data analysis for two real life datasets are illustrated.

Maximum likelihood estimation for multi-factor uncertain differential equations

Multi-factor uncertain differential equation (MUDE) is widely used in dynamic systems with multiple types of uncertainties. After observing actual data, how to estimate unknown parameters based on these observations rationally is a key issue. Existing estimation methods are based on difference scheme of MUDE with a prerequisite that observation intervals are small enough. To overcome this shortage, a novel maximum likelihood method is proposed using the concept of residual for MUDE in this work. Both analytical and numerical methods for calculating residuals are also studied. In addition, uncertain hypothesis test is employed to verify the reasonableness of estimations. Numerical example and real data analyses show the proposed method in details.

Multivariate Poisson cokriging: A geostatistical model for health count data.

Multivariate disease mapping is important for public health research, as it provides insights into spatial patterns of health outcomes. Geostatistical methods that are widely used for mapping spatially correlated health data encounter challenges when dealing with spatial count data. These include heterogeneity, zero-inflated distributions and unreliable estimation, and lead to difficulties when estimating spatial dependence and poor predictions. Variability in population sizes further complicates risk estimation from the counts. This study introduces multivariate Poisson cokriging for predicting and filtering out disease risk. Pairwise correlations between the target variable and multiple ancillary variables are included. By means of a simulation experiment and an application to human immunodeficiency virus incidence and sexually transmitted diseases data in Pennsylvania, we demonstrate accurate disease risk estimation that captures fine-scale variation. This method is compared with ordinary Poisson kriging in prediction and smoothing. Results of the simulation study show a reduction in the mean square prediction error when utilizing auxiliary correlated variables, with mean square prediction error values decreasing by up to 50%. This gain is further evident in the real data analysis, where Poisson cokriging yields a 74% drop in mean square prediction error relative to Poisson kriging, underscoring the value of incorporating secondary information. The findings of this work stress on the potential of Poisson cokriging in disease mapping and surveillance, offering richer risk predictions, better representation of spatial interdependencies, and identification of high-risk and low-risk areas.

Influencer detection meets network autoregression — Influential regions in the bitcoin blockchain

Known as an active global virtual money network, the Bitcoin blockchain, with millions of accounts, has played a continually increasingly important role in fund transition, digital payment, and hedging. We propose a method to Detect Influencers in Network AutoRegressive models (DINAR) via sparse-group regularization to detect regions influencing others across borders. For a granular analysis, we analyse whether the transaction size plays a role in the dynamics of the cross-border transactions in the network. With two-layer sparsity, DINAR enables discovering (1) the active regions with influential impact on the global digital money network and (2) whether changes in the size of the transaction affect the dynamic evolution of Bitcoin transactions. In the analysis of real data of the Bitcoin blockchain from Feb 2012 to December 2021, we find that influence from certain regions is linked to the economic need to use BTC, such as to circumvent sanctions, avoid high inflation, and to carry out transactions through off-shore markets. The effects are robust to different groupings, evaluation periods, and choices of regularization parameters.

Variance-reduced sampling importance resampling

The sampling importance resampling method is widely utilized in various fields, such as numerical integration and statistical simulation. In this paper, two modified methods are presented by incorporating two variance reduction techniques commonly used in Monte Carlo simulation, namely antithetic sampling and Latin hypercube sampling, into the process of sampling importance resampling method, respectively. Theoretical evidence is provided to demonstrate that the proposed methods significantly reduce estimation errors compared to the original approach. Furthermore, the effectiveness and advantages of the proposed methods are validated through both numerical studies and real data analysis.

Some asymptotic results of a kNN conditional mode estimator for functional stationary ergodic data

. This article investigates the conditional mode estimation of a univariate response variable given a functional random covariate (i.e., valued in some infinite-dimensional space) whenever a stationary ergodic data are considered. We construct a new estimator of the conditional mode which is constructed by a kNN approach. Under less restrictive conditions, we show the strong consistency of the proposed estimator. To assess the efficiency of the developed estimator, empirical analysis as well as real data analyses are performed.

Genome-wide fine-mapping improves identification of causal variants.

Fine-mapping refines genotype-phenotype association signals to identify causal variants underlying complex traits. However, current methods typically focus on individual genomic segments without considering the global genetic architecture. Here, we demonstrate the advantages of performing genome-wide fine-mapping (GWFM) and develop methods to facilitate GWFM. In simulations and real data analyses, GWFM outperforms current methods in error control, mapping power and precision, replication rate, and trans-ancestry phenotype prediction. For 48 well-powered traits in the UK Biobank, we identify causal variants that collectively explain 17% of the SNP-based heritability, and predict that fine-mapping 50% of that would require 2 million samples on average. We pinpoint a known causal variant, as proof-of-principle, at FTO for body mass index, unveil a hidden secondary variant with evolutionary conservation, and identify new missense causal variants for schizophrenia and Crohn's disease. Overall, we analyse 600 complex traits with 13 million SNPs, highlighting the efficacy of GWFM with functional annotations.