Browsing by Author "Zhao, Yi"
- Results Per Page
- Sort Options
Item Open Access Bayesian Multiregression Dynamic Models with Applications in Finance and Business(2015) Zhao, YiThis thesis discusses novel developments in Bayesian analytics for high-dimensional multivariate time series. The focus is on the class of multiregression dynamic models (MDMs), which can be decomposed into sets of univariate models processed in parallel yet coupled for forecasting and decision making. Parallel processing greatly speeds up the computations and vastly expands the range of time series to which the analysis can be applied.
I begin by defining a new sparse representation of the dependence between the components of a multivariate time series. Using this representation, innovations involve sparse dynamic dependence networks, idiosyncrasies in time-varying auto-regressive lag structures, and flexibility of discounting methods for stochastic volatilities.
For exploration of the model space, I define a variant of the Shotgun Stochastic Search (SSS) algorithm. Under the parallelizable framework, this new SSS algorithm allows the stochastic search to move in each dimension simultaneously at each iteration, and thus it moves much faster to high probability regions of model space than does traditional SSS.
For the assessment of model uncertainty in MDMs, I propose an innovative method that converts model uncertainties from the multivariate context to the univariate context using Bayesian Model Averaging and power discounting techniques. I show that this approach can succeed in effectively capturing time-varying model uncertainties on various model parameters, while also identifying practically superior predictive and lucrative models in financial studies.
Finally I introduce common state coupled DLMs/MDMs (CSCDLMs/CSCMDMs), a new class of models for multivariate time series. These models are related to the established class of dynamic linear models, but include both common and series-specific state vectors and incorporate multivariate stochastic volatility. Bayesian analytics are developed including sequential updating, using a novel forward-filtering-backward-sampling scheme. Online and analytic learning of observation variances is achieved by an approximation method using variance discounting. This method results in faster computation for sequential step-ahead forecasting than MCMC, satisfying the requirement of speed for real-world applications.
A motivating example is the problem of short-term prediction of electricity demand in a "Smart Grid" scenario. Previous models do not enable either time-varying, correlated structure or online learning of the covariance structure of the state and observational evolution noise vectors. I address these issues by using a CSCMDM and applying a variance discounting method for learning correlation structure. Experimental results on a real data set, including comparisons with previous models, validate the effectiveness of the new framework.
Item Open Access Characterization of Gene-by-Age Interaction and Gene-by-Gene Interaction In Coronary Artery Disease(2012) Zhao, YiThe success of genome-wide association studies (GWAS) has been limited by missing heritability and lack of biological relevance of identified variants. We sought to address these issues by characterizing interaction among genotypes and environment using case-control samples enrolled at Duke University Medical Center. First, we studied the impact of age on coronary artery disease (CAD). Gene-by-age (GxAGE) interactions were tested at genome-wide scale, along with genes' marginal effects in age-stratified groups. Based on the interaction model, age plays the role as a modifier of the age-CAD relationship. SNPs associated with CAD in both young and old demonstrate consistency in effect sizes and directions. In spite of these SNPs, vastly different CAD associated genes were discovered across age and race groups, suggesting age-dependent mechanisms of CAD onset. Second, we explored gene-by-gene interaction (GxG) using a statistical model and compared results to biological evidence. Specifically, we investigated GATA2 as a candidate gene transcription factor, and modeled the interaction with genome-wide SNPs. The genetic effects at interacting loci were modified by GATA2 genotype. Without taking GATA2 variants into account , no marginal main effects were detected. Open access ChIP-seq data was available for comparison with the statistical model, and to relate GWAS findings with biological mechanisms. The agreement between the statistical and biological models was very limited.
Item Open Access Dietary Patterns, Nutritional Status, Prevalence and Risk Factors for Anemia among School Children in Naama Community, Uganda(2015) Zhao, YiBackground: The disease burden of malnutrition, anemia, malaria and helminth infections among school-aged children is rarely studied in low- and middle-income countries (LMICs) although these children are still at a high risk for these diseases compared with other more studied populations, such as pregnant women and children under 5 years old. Even in countries where the prevalence and trend of anemia and malnutrition are relatively well documented, few studies relate this to dietary factors, which are considered major contributors to anemia and malnutrition in most age groups. Aims: The aims of the study are (1) To determine the prevalence of anemia, malaria, helminth infections and malnutrition in a sample of 95 children, ages 6 to 14, attending primary schools in Naama Community, Uganda, and to observe and quantify school children's dietary patterns and daily nutrient intakes. (2) To find out the association between dietary factors and the risk of anemia and malnutrition. Methods: Measures included school-based, cross-sectional surveys, dietary assessments, anthropometric measurements and biological tests among school children. Photo-assisted 24-hour recall was used to collect daily nutrient intakes, combined with a Food Frequency Questionnaire (FFQ) to capture the dietary patterns. Anthropometrical and biochemical data was collected using standardized protocols. Socioeconomic data was obtained from parent surveys. Results: Ninety-five children in total were enrolled in all or some components of the study. The prevalence of anemia was low (3.2%), and all were mildly anemic. However, the prevalence of malaria and hookworm infections was relatively high, representing 12.9% and 24.4% of the studied population, respectively. In the studied children, 2.8% were underweight, 15.6% stunted and 1.3% thin, using criteria based on the WHO Growth reference. According to the WHO recommendations for nutrient intake, 80% of participants consumed inadequate energy from their daily diet, especially boys. Dietary fat intake was insufficient in 78% of the children. About 25% of the children had a low protein intake and 93% had low intakes of vitamin A. Calcium intake was low in school children's diet- 97.6% of children lacked of calcium. Inadequate vitamin C intake was less common, appearing in one out of three participants (29.1%). Matooke and posho, the most common local staple food, were the major sources for children's energy. Avocado, beans and matooke contributed to the highest fat, protein and vitamin A intake, respectively. Every participant reported consuming cereal/cereal products, roots/tubers/plantain, pulses/nuts, oil/oil-rich foods and fruits during the previous month, while almost one-fifth never consumed milk/dairy products. Vegetables were consumed by most of the respondents. Few (1.6%) of the respondents reported no animal source food in their diet. In general, children ate 3.81(SD: 0.99) out of all five meals in the area. Dinner was consumed by most of the participants (93.7%) while afternoon tea was the least consumed meal and was skipped by half of the childrenNo significant association between dietary factors and anemia were found, mainly due to the small sample size and low prevalence of disease. None of the dietary factors of primary interest were found to be associated with children's Hb concentration, but secondary analysis found the frequency of eating oranges was a protective factor for higher Hb concentration (P=0.015). The association between diet and stunting was not significant, except children who had low dietary fat intakes had a lower risk of being stunted compared with those had adequate fat consumption (OR=0.27, P=0.046). Conclusions: Malaria and helminth infections, but not anemia, in this cohort of school children is relatively high. Stunting was the most prevalent type of malnutrition. Most of the nutrients studied were not adequate in children's diets. The dietary pattern in this sample of children was primarily high in carbohydrates from staple foods and a minimal intake of fat and protein from animal sources. Children generally ate four meals per day. Dietary factors do not explain anemia and stunting in this population.