Masters Theses

Permanent URI for this collection

https://jewlscholar.mtsu.edu/handle/mtsu/3482

Browse

Browsing Masters Theses by Department "Basic & Applied Sciences"
  • Item
    A Comparative Study on Two Strategies for Distributed Classification
    (Middle Tennessee State University, 2018-05-30) Xu, Honglan ; Wu, Qiang ; Hong, Don ; Liu, Yeqian ; Green, Lisa ; Basic & Applied Sciences
    Distributed learning is an effective tool to process big data. An easy and effective distributed learning approach is the divide and conquer method. It first partitions the whole data set into multiple subsets. A base learning algorithm is then applied to each subset. Finally the results from these subsets are coupled together. In the classification setting, many classification algorithms can be used in the second stage. Typical ones include the logistic regression and support vector machines. For the third stage, both voting and averaging can be used as the coupling strategies. In this thesis, empirical studies are done to thoroughly compare the effectiveness of these two coupling strategies. Averaging is found to be more effective in most scenarios.
  • Item
    A Duality between hypergraphs and cone lattices
    (Middle Tennessee State University, 2018-03-22) French, Zack ; Hart, James ; Sarkar, Medha ; Ye, Dong ; Basic & Applied Sciences
    In this paper, we introduce and characterize the class of lattices that arise as the
  • Item
    Analysis of MTSU Student Retention Data
    (Middle Tennessee State University, 2016-03-23) Baghernejad, Danielle Marcella ; Wu, Qiang ; Calahan, Rebecca ; Green, Lisa ; Li, Cen ; Basic & Applied Sciences
    Student retention is a challenging task in higher education, since in general more students remaining in the university means better academic programs and higher revenue. Thus, improving retention rates can not only help current students achieve academic success, but help future students as well. The objective of this thesis is to employ data mining and predictive tools on student data to predict student retention among the freshman students. In particular, we aim to identify freshman students who are more likely to drop out so that preemptive actions can be taken by the university. Through data analysis, relevant variables are identified to incorporate into models for prediction. Missing values are taken into consideration, and missing value imputation methods are explored.
  • Item
    BIFURCATION ANALYSIS IN APOPTOSIS (RECEPTOR CLUSTERING)
    (Middle Tennessee State University, 2018) Spears, Genesis Amelia ; Basic & Applied Sciences
    Apoptosis is a designed cell death mechanism involved in biological processes. Apoptosis can either be activated by extrinsic pathway or by the intrinsic pathway. A major part of the external apoptosis pathway is the death receptor Fas which, on binding to its associated ligand FasL, they eventually form the death-inducing signaling complex. FasL promotes clustering for open Fas and activates open stable Fas, forming locally stable signaling platforms through neighborhood-induced receptor interactions. The model exhibits a bifurcation called hysteresis, providing an upstream mechanism for bistability and robustness to decide if the cell lives or dies. At low receptor concentrations, the bistability depends on three states of FasL. The irreversible bistability, representing a committed cell death decision, emerges at high receptor concentrations. Furthermore, the model suggests a mechanism by which cells may function as bistable life/death switches which are independent of their downstream dynamic components. This will be illustrated by simulations.
  • Item
    Collective Behavior Modeling Through Velocity Alignment
    (Middle Tennessee State University, 2019) Zanussi, Jacy Thor ; Sinkala, Zachariah ; Ding, Wandi ; Khaliq, Abdul ; Basic & Applied Sciences
    We construct a model for collective behavior phenomena by undermining the assumption that the rate of change of position equals velocity in the particle Cucker-Smale model for flocking. Conditions for collective behavior are proven and three continuous models for segregation are presented with simulations for two of them. Future avenues of research and a variety of applications such as social science, engineering, and business trends are discussed in the conclusion.
  • Item
    Data Mining and Machine Learning Algorithms for Workers' Compensation Early Severity Prediction
    (Middle Tennessee State University, 2016-06-21) Mathews, David ; Hong, Don ; Wu, Qiang ; Green, Lisa ; Hart, James ; Basic & Applied Sciences
    Although the number of workers' compensation claims have been declining over the last two decades, average cost per claim has been steadily increasing. Identifying factors that contribute to severe claims and effectively managing those claims early in the claim life-cycle could reduce costs for employers and insurers. This research project utilizes machine learning algorithms to predict a binary severity outcome variable. A text mining algorithm, Correlated Topics Model, was used to convert textual description fields to topics. Support Vector Machines and Regularized Logistic Regression were implemented for severity classification and variable selection, respectively. Due to the asymmetric severity outcomes in the training data, a balancing method for matching the volume of severe/non-severe claims was employed. Optimal model parameters for both algorithms were selected based on a profitability metric and 10-fold cross-validation. Discussion of data processing techniques and mathematical exposition of machine learning algorithms are provided. Open source statistical programming software, R, was utilized in this project.
  • Item
    Decomposition of Cubic Graphs on the Torus and Klein Bottle
    (Middle Tennessee State University, 2015-10-30) Bachstein, Anna Caroline ; Ye, Dong ; Zha, Xiaoya ; Stephens, David ; Basic & Applied Sciences
    It was conjectured by Hoffman-Ostenhof that the edge set of every cubic graph can be decomposed into a spanning tree, a matching, and a family of cycles. This conjecture was verified for many graphs such as the Peterson graph, prisms over cycles, and Hamiltonian graphs. Later the conjecture was also verified for 3-connected cubic graphs on the plane and protective plane by Kenta Ozeki and Dong Ye. In this paper we will verify the conjecture for 3-connected cubic graph on the torus and Klein bottle.
  • Item
    Distributed Computing Approaches to Pathfinding Problems
    (Middle Tennessee State University, 2016-06-19) Myers, Robert Vital ; Phillips, Joshua ; Gu, Yi ; Barbosa, Salvador ; Basic & Applied Sciences
    The problem of determining the existence of a path between vertices in problem domains with large graphs is outpacing the increases in commonly available processor speeds. This presents a growing need for pathfinding algorithms which can capitalize on parallel approaches. These approaches are often based on parallelizing the search on a single machine. However, some problems may be so large that it becomes appropriate to use distributed computing. This research explores the Distributed Fringe Search algorithm as a more conducive approach for pathfinding problems over multiple distributed machines. The work presented here is novel in its extension of DFS by developing the Distributed Computing Fringe Search. Additionally, this research proposes the Hash Distributed Fringe Search that utilizes space abstraction techniques for work distribution and a more uniform memory requirement. Finally, results are presented to show the impact of the approaches in large searches; these results inform suggestions for future work.
  • Item
    THE EFFECT OF HAY TYPE AND SOAKING ON GLYCEMIC RESPONSE IN HORSES
    (Middle Tennessee State University, 2015-04-09) Collins, Josie Lynn ; Hoffman, Rhonda ; Spooner, Holly ; Haffner, John ; Basic & Applied Sciences
    Dietary management of equine metabolic syndrome includes minimizing glycemic response. Soaking hay prior to feeding reduces nonstructural carbohydrate content, but little research indicates if soaking hay reduces glycemic response. Glycemic response of four hay diets were evaluated: dry or soaked prairiegrass hay, and dry or soaked alfalfa hay. Twelve healthy horses were randomly assigned into two groups and fed the hay diets at 0.5% BW in a 2x2 factorial design. Blood samples were collected at 0, 30, 60, 90, 120, 180, 240, and 300 min after feeding. Plasma glucose was analyzed using a colorimetric assay, and incremental area under the curve (AUC) of glucose response calculated. Data were analyzed using a mixed model with repeated measures. Plasma glucose and the AUC was higher (P =0.0001) in healthy horses fed alfalfa compared to grass hay, with no differences identified due to soaking (P = 0.82). Additional research is needed to determine if soaking hay has physiological merit in horses with metabolic issues.
  • Item
    EFFECTS OF WHOLE BODY VIBRATION ON LAMENESS, STRIDE LENGTH, CORTISOL, AND OTHER PARAMETERS IN HEALTHY HORSES
    (Middle Tennessee State University, 2018-04-13) Sugg, Seneca ; Spooner, Holly ; Hoffman, Rhonda ; Haffner, John ; Basic & Applied Sciences
    Whole body vibration (WBV) is a popular treatment modality in the horse industry anecdotally believed to improve performance, yet little research exists. This study examined the effect of WBV on stride and lameness parameters, as well as salivary cortisol and heart rate. Six treatment horses (VIB) were vibrated at 50 Hz for 45 minutes 6 days per week and compared to stalled controls (CON) (Aim 1). Further, VIB was analyzed for parameters at different time periods immediately before treatment (pre), after (post), and 1, 2, 4, and 24 hours after (Aim 2). For Aim 1, minimum pelvis displacement was lower in VIB versus CON (P < .05), suggesting more hind end stability. Both groups decreased in stride length from d 0 to d 14 or d 28 (P < 0.01). For Aim 2, no differences were found in lameness or stride length at any time point. However, salivary cortisol concentrations were lower after (P < 0.05) and heart rate showed a similar pattern (P = 0.06). Whole body vibration appears to have an acute relaxation effect in stalled, healthy horses.
  • Item
    Factors that influence the presence of fecal indicator bacteria from three potential exposure pathways
    (Middle Tennessee State University, 2019) Stallard, Megan ; Bailey, Frank C. ; Farone, Mary ; Fischer, Bud ; Otter, Ryan R. ; Wang, Sherry ; Basic & Applied Sciences
    Fecal indicator bacteria (FIB) are good predictors for the presence of pathogens associated with fecal contamination in recreational waters and criteria have been created as a benchmark to assess risk to human health. The two most traditionally used FIB for this purpose are culturable Escherichia coli in freshwaters and enterococci in marine settings and members of the Order Bacteroidales have been mentioned heavily in the literature as supplementary indicators. It has been well documented in the literature that fecal indicator bacteria can be modulated by a number of factors, such as temperature, ultraviolet light, land use, and rainfall. This collection of studies yielded information on a how factors could modulate FIB from a variety of transport pathways to human exposure. In surface waters, concentration- and loading-based results for E. coli and Bacteroidales were highest in summer and spring, and lowest in the winter and fall, respectively. Bacteroidales concentrations were positively correlated with temperature and total suspended solids and negatively correlated with dissolved oxygen. In beach sand, E. coli concentrations were highest in the upper 0-10 cm of the foreshore samples where beachgoers typically congregate. For substrate types typically used in stormwater infrastructure, plate count concentrations increased considerably from initial spiking dose, but results were dependent on strain of bacteria and substrate type (concrete, metal, PVC).
  • Item
    GARDNERELLA VAGINALIS STRAIN DEPENDENT GROWTH AND CYTOTOXICITY IN HUMAN MONOCYTES
    (Middle Tennessee State University, 2018) Ouellette, Corbett Charles ; Basic & Applied Sciences
    Bacterial vaginosis (BV) is the most prevalent vaginal infection for women of reproductive age, characterized by a lack of normal commensal bacteria and a proliferation of pathogenic species. BV infections can lead to sterility, pelvic inflammatory disease, increased chance of HIV transmission, and preterm delivery. Gardnerella vaginalis has been found in many cases of BV, although G. vaginalis has been isolated in women exhibiting no diagnostic criteria. The goal of this study was to examine the growth kinetics and cytotoxicity of different Gardnerella vaginalis strains, which may provide important information in identifying growth conditions of asymptomatic BV. G. vaginalis comparison studies utilized strains 14018, 14019, and 49145, and were grown in pure bacterial culture in brain-heart infusion (BHI) and in Roswell Park Memorial Institute (RPMI) medium, or co-cultured with undifferentiated or differentiated human monocyte cells (THP-1) in RPMI. Significant differences in growth were noted among all strains in BHI and RPMI (p < 0.05). Further experiments showed that there is a significant difference in the cytotoxicity of the strains when in co-culture with THP-1 cells (p < 0.01), and may be attributed to either different growth patterns and metabolic proteins, or possibly due to further genetic differences found among virulence factors. In silico analysis of the three strains was done to provide supporting information relevant to the growth differences. The results of this study may contribute to a better understanding of how the genetic differences found within the different strains of Gardnerella vaginalis correlate with differences BV.
  • Item
    HIGH SCHOOL STATISTICS TEACHERS’ UNDERSTANDING OF HYPOTHESIS TESTING THROUGH SIMULATION
    (Middle Tennessee State University, 2018) Matuszewski, Amber L. ; Basic & Applied Sciences
    It is a growing trend in statistics education to use simulations for hypothesis testing due to the belief that simulations help make the abstract ideas behind hypothesis testing become more concrete and understandable. Using simulations involves randomizing data production, repeating by simulation to see what is typical, and rejecting the null hypothesis if your data falls in the tails of the simulated distribution. This explanatory multiple case study sought to answer the following questions, “How does engaging in simulation tasks for hypothesis testing influence high school statistics teachers’ understanding of traditional hypothesis testing?” and “How do simulation tasks influence high school statistics teachers’ understanding of simulations and how do they make connections between traditional and simulation approaches for hypothesis testing?” The results of this study revealed that teachers’ understanding of hypothesis testing was positively impacted because of engaging in simulation tasks. The focus of the simulation tasks on concepts and logic instead of procedures helped the participants develop their understanding of the logic of hypothesis testing. Additionally, simulation approaches focus on visualizations, which helped develop the understanding of the probabilistic nature of hypothesis testing. Finally, how teachers understood simulations and made connections between traditional and simulation approaches was directly influenced by the lesson plan design.
  • Item
    IN-WHEEL HUB SRM SIMULATION AND ANALYSIS
    (Middle Tennessee State University, 2014-03-28) Sager, Milton ; Melnikov, Yuri ; Ding, Wandi ; Sbenaty, Saleh ; Basic & Applied Sciences
    Is it feasible to replace the conventional gasoline engine and subsequent drive sys- tem in a motorcycle with an electric switched reluctance motor (SRM) by placing the SRM inside the rear wheel, thereby removing the need for things such as a clutch, chain, transmission, gears and sprockets?
  • Item
    INFLUENCE OF WHOLE BODY VIBRATION ON BONE DENSITY IN THE STALLED HORSE
    (Middle Tennessee State University, 2015-04-10) Hulak, Emily Sarah ; Spooner, Holly ; Hoffman, Rhonda ; Brzezicki, Anne ; Basic & Applied Sciences
    High frequency mechanical accelerations resulting from whole body vibration (WBV) have a strong osteogenic effect. It is hypothesized that WBV will maintain bone density in stalled horses equal to that of a stalled control group receiving light exercise. Radiographs of the third metacarpal were taken on d 0, 28, 56 of a stalled control receiving light exercise (n=6), and a stalled treatment group receiving WBV (n=6) for determination of bone mineral content (BMC) via radiographic bone aluminum equivalence (RBAE). No differences were observed in BMC due to treatment in the medial (P = 0 .98), lateral (P = 0.93), dorsal (P = 0.69), or palmar (P = 0.90) cortices. These results suggest that this level of WBV did not increase BMC, but maintained a baseline level similar to that of a horse in low intensity exercise, thus vibration therapy should be considered for horses subjected to stalling without exercise.
  • Item
    ISOPERIMETRIC CONSTANTS IN PLANAR GRAPHS WITH HYPERBOLIC PROPERTIES
    (Middle Tennessee State University, 2014-05-25) Whitlatch, Hays Wimsatt ; Zha, Xiaoya ; Hart, James ; Nelson, Don ; Stephens, Chris ; Ye, Dong ; Basic & Applied Sciences
    Isoperimetric inequalities date back to ancient Greece where figures with equal perimeters but different shapes were compared (Zenodorus, On Isoperimetric Figures). The original problem was to maximize the area contained within a curve of specified length. In Euclidean geometry the result is a circle. This can be generalized to shapes on non-Euclidean surfaces as well as for higher dimensions where we seek to maximize the hyperdimensional volume respective to the hyperdimensional surface area.
  • Item
    Lattice Structures in Finite Graph Topologies
    (Middle Tennessee State University, 2015-08-15) Frazier, Brian Phillip ; Hart, James ; Ye, Dong ; Stephens, Chris ; Seo, Suk ; Basic & Applied Sciences
    In his 2005 dissertation, Antoine Vella studied the relationships between hypergraphs
  • Item
    Medical Trend Analysis Methods
    (Middle Tennessee State University, 2014-12-31) Yin, Le ; Wu, Qiang ; Hong, Don ; Sinkala, Zachariah ; Basic & Applied Sciences
    Medical trend is the most important component used to indicate and file rates. Insurance companies use trend to forecast future costs and premiums. Governments use medical trend in the rate review process. This thesis reviews four methods used to find a trend factor: average ratio, linear regression, exponential regression and time series analysis method with rolling average technology. A software package is developed to calculate medical trend based on annual data or monthly data. An efficient method to detect the outliers is also presented.
  • Item
    New Algorithms for Supervised Dimension Reduction
    (Middle Tennessee State University, 2019) Zhang, Ning ; Wu, Qiang ; Hong, Don ; Li, Cen ; Robertson, William ; Basic & Applied Sciences
    Advances in data collection and storage capabilities during the past decades have led to information overload in most sciences and ushered in a big data era. Data of big volume, as well as high dimensionality, become ubiquitous in many scientific domains. They present many mathematical challenges as well as some opportunities and are bound to give rise to new theoretical developments. Dimension reduction aims to explore low dimensional representation for high dimensional data. It helps promote the understanding of the data structure through visualization and enhance the predictive performance of machine learning algorithms by preventing the “curse of dimensionality.” As high dimensional data become ubiquitous in modern sciences, dimension reduction methods are playing more and more important roles in data analysis. The contribution of this dissertation is to propose some new algorithms for supervised dimension reduction that can handle high dimensional data more efficiently. The first new algorithm is the overlapping sliced inverse regression (OSIR). Sliced inverse regression (SIR) is a pioneer tool for supervised dimension reduction. It identifies the subspace of significant factors with intrinsic lower dimensionality, specifically known as the effective dimension reduction (EDR) space. OSIR refines SIR through an overlapping slicing scheme and can estimate the EDR space and determine the number of effective factors more accurately. We show that the overlapping procedure has the potential to identify the information contained in the derivatives of the inverse regression curve, which helps to explain the superiority of OSIR. We prove that OSIR algorithm is √n-consistent. We also propose the use of bagging and bootstrapping techniques to further improve the accuracy of OSIR. Online learning has attracted great attention due to the increasing demand for systems that have the ability of learning and evolving. When the data to be processed is also high dimensional, and dimension reduction is necessary for visualization or prediction enhancement, online dimension reduction will play an essential role. We propose four new online learning approaches for supervised dimension reduction, namely, the incremental sliced inverse regression, the covariance-free incremental sliced inverse regression, the incremental overlapping sliced inverse regression, and the covariance-free incremental overlapping sliced inverse regression. All four methods are able to update the EDR space fast and efficiently when new observations come in. The effectiveness and efficiency of all four algorithms are verified by simulations and real data applications.
  • Item
    Plasma and muscle amino acid concentrations in insulin resistant compared to normal horses in the fed and fasted state
    (Middle Tennessee State University, 2017-04-05) Macon, Erica Lyn ; Hoffman, Rhonda ; Haffner, John ; Spooner, Holly ; Basic & Applied Sciences
    The effects of insulin resistance on protein and amino acid metabolism have not been documented in horses, despite knowledge that insulin stimulates the incorporation of amino acids into proteins and decreases the rate of protein degradation in skeletal muscle. The objective of this research was to investigate protein metabolism in insulin resistant (IR) horses compared to Normal controls. Preceding the study, horses were screened for fasting plasma insulin concentrations with insulin concentrations of 10.1 3.5 mU/L = Normal, and 41.2 15.9 mU/L = IR, and then underwent an oral sugar challenge (OST). Blood samples were collected before and at 60, 75, 90, 120, 180 min after administration of Karo Light Syrup (0.15 mL/kg of BW) for the determination of plasma insulin and glucose. Based on the OST, eight horses (16 + 3 yrs), 4 IR and 4 Normal, were studied while receiving Purina Strategy and a mixture of Timothy/Bermudagrass hay fed at 2% BW, a diet which met or exceeded NRC recommendations. After a week of dietary adaption, the morning meal (half the daily ration) was fed on day 7. Blood samples were taken at 0, 1, 3, 4, and 6 h post-prandial for determination of plasma amino acids. On day 8, muscle biopsies were taken and immediately frozen in liquid nitrogen for analysis of free amino acids. Glucose was analyzed using an enzymatic assay, and insulin by radioimmunoassay. Plasma free amino acids and semitendinosus muscle free amino acids were determined using reverse-phase HPLC of phenylisothiocyanate derivatives. Data were analyzed using a mixed model with repeated measures analysis of SAS, with time and group as main effects. The OST data confirmed higher insulin (P = 0.020) and a trend for higher glucose (P = 0.055) in IR vs Normal horses. There were no differences between IR and Normal for any plasma free amino acids (P > 0.15) or semitendinosus muscle free amino acids (P > 0.17). Contradictory to this study, hyperinsulinemic clamp procedures in healthy horses and pigs lowered plasma amino acid concentrations, with similar results reported in healthy and diabetic humans. Lack of variation in amino acid concentrations between IR and Normal horses suggests that insulin resistance does not affect amino acid absorption into the plasma pool or incorporation into the muscle in horses.