College of Agriculture & Natural Resources

Permanent URI for this communityhttp://hdl.handle.net/1903/1598

The collections in this community comprise faculty research works, as well as graduate theses and dissertations.

Browse

Subject: search.filters.subject.Statistics

Search Results

Now showing 1 - 7 of 7
  • Thumbnail Image
    Item
    A COMPARATIVE ANALYSIS OF RANDOM FOREST AND LOGISTIC REGRESSION FOR WEED RISK ASSESSMENT
    (2018) Harris, Chinchu; Peer, Wendy; Plant Science and Landscape Architecture (PSLA); Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Invasive species have largely negative impacts on the environment and the economy. The management and regulation of invasive plants are facilitated using screening tools, such as weed risk assessments (WRAs) to predict the invasive potential of non-native plants. The identification of these species and their subsequent regulation on importation helps to reduce the risk of future ecosystem and economic costs. Globally, there are many different types of highly useful WRAs already available. However, in this day of big data and powerful predictive analytics, there is an increasing demand for the development of new and more robust screening tools. In this thesis, I use the machine learning algorithm, Random forests, to develop a new WRA. I show that random forest model has greater predictive accuracies than an existing logistic regression model and that random forest is a better learner. In addition, variable importance analysis was performed to identify factors associated with invasive status classification of non-native plants. The study suggests that random forests make powerful weed risk screening tools and should be utilized for assessing invasive risk potential along with other WRAs. An integrative approach for evaluating weed risk can greatly serve to facilitate the WRA process.
  • Thumbnail Image
    Item
    THE ROLE OF SOILS IN PRODUCTION: AGGREGATION, SEPARABILITY, AND YIELD DECOMPOSITION IN KENYAN AGRICULTURE
    (2015) Pieralli, Simone; Chambers, Robert G; Agricultural and Resource Economics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Agricultural production relies on soils. Increasing global population and the impact of climate change threaten the sustainability of soil for agricultural production. For these reasons, it is necessary to broaden present current methodological approaches to incorporating soil into economic analysis. The first essay proposes a methodology to aggregate quantitative soil characteristics through the use of separability theory in a Data Envelopment Analysis framework. This yields an aggregate soil-quality measure that appropriately aggregates soil characteristics. The application is to Kenyan maize farmers. The second essay develops a nonparametric statistical test of structural separability based on a bias correction of a central limit theorem for Data Envelopment Analysis estimators developed in Kneip, Simar and Wilson (2015a). The proposed nonparametric test for structural separability adapts the statistical procedures to test technology restrictions present in Kneip, Simar and Wilson (2015b). Monte Carlo experiments determine the size and power properties of the proposed test. An empirical analysis of Kenyan household farmers illustrates the use of the methodology. Global needs for higher agricultural production require understanding whether the frequently noted inverse land size-yield relationship is a true empirical regularity or an artifact of data collection methods. To examine this relationship, the third essay of this dissertation generalizes productivity decomposition methods to incorporate the quantification of a soil-productivity contribution. The generalized method decomposes a yield index into separate components attributable to (1) efficiency, (2) soil quality, (3) land size, (4) variable inputs, (5) capital inputs, and (6) output mix. Nonparametric productivity accounting methods are used to decompose the inverse land size-yield relationship in a multi-output representation of the technology without specific assumptions on returns to scale. A strongly significant inverse land size-yield relationship is present among Kenyan farmers.
  • Thumbnail Image
    Item
    Implications of heterogeneity in discrete choice analysis
    (2013) Martinez-Cruz, Adan Leobardo; McConnell, Kenneth E.; Agricultural and Resource Economics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    This dissertation carries out a series of Monte Carlo simulations seeking the implications for welfare estimates from three research practices commonly implemented in empirical applications of mixed logit and latent class logit. Chapter 3 compares welfare measures across conditional logit, mixed logit, and latent class logit. The practice of comparing welfare estimates is widely used in the field. However, this chapter shows comparisons of welfare estimates seem unable to provide reliable information about the differences in welfare estimates that result from controlling for unobserved heterogeneity. The reason is that estimates from mixed logit and latent class logit are inherently inecient and inaccurate. Researchers tend to use their own judgement to select the number of classes of a latent class logit. Chapter 4 studies the reliability of welfare estimates obtained under two scenarios for which an empirical researcher using his/her judgement would arguably choose less classes than the true number of classes. Results show that models with a number of classes smaller than the true number tend to yield down- ward biased and inaccurate estimates. The latent class logit with the true number of classes always yield unbiased estimates but their accuracy may be worse than models with the smaller number of classes. Studies implementing discrete choice experiments commonly obtain estimates of preference parameters from latent class logit models. This practice, however, implies a mismatch: discrete choice experiments are designed under the assumption of homogeneity in preferences, and latent class logit search for heterogeneous preferences. Chapter 5 studies whether welfare estimates are robust to this mismatch. This chapter checks whether the number of choice tasks impact the reliability of welfare estimates. The findings show welfare estimates are unbiased regardless the number of choice tasks, and their accuracy increases with the number of choice tasks. However, some of the welfare estimates are inefficient to the point that cannot be statistically distinguished from zero, regardless the number of choice tasks. Implications from these findings for the empirical literature are discussed.
  • Thumbnail Image
    Item
    Genome-Wide Analysis of Histone Modification Enrichments Induced by Marek's Disease Virus in Inbred Chicken Lines
    (2013) Mitra, Apratim; Song, Jiuzhou; Animal Sciences; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Covalent histone modifications constitute a complex network of transcriptional regulation involved in diverse biological processes ranging from stem cell differentiation to immune response. The advent of modern sequencing technologies enables one to query the locations of histone modifications across the genome in an efficient manner. However, inherent biases in the technology and diverse enrichment patterns complicate data analysis. Marek's disease (MD) is an acute, lymphoma-inducing disease of chickens with disease outcomes affected by multiple host and environmental factors. Inbred chicken lines 63 and 72 share the same major histocompatibility complex haplotype, but have contrasting responses to MD. This dissertation presents novel methods for analysis of genome-wide histone modification data and application of new and existing methods to the investigation of epigenetic effects of MD on these lines. First, we present WaveSeq, a novel algorithm for detection of significant enrichments in ChIP-Seq data. WaveSeq implements a distribution-free approach by combining the continuous wavelet transform with Monte Carlo sampling techniques for effective peak detection. WaveSeq outperformed existing tools particularly for diffuse histone modification peaks demonstrating that restrictive distributional assumptions are not necessary for accurate ChIP-Seq peak detection. Second, we investigated latent MD in thymus tissues by profiling H3K4me3 and H3K27me3 in infected and control birds from lines 63 and 72. Several genes associated with MD, e.g. MX1 and CTLA–4, along with those linked with human cancers, showed line-specific and condition-specific enrichments. One of the first studies of histone modifications in chickens, our work demonstrated that MD induced widespread epigenetic variations. Finally, we analyzed the temporal evolution of histone modifications at distinct phases of MD progression in the bursa of Fabricius. Genes involved in several important pathways, e.g. apoptosis and MAPK signaling, and various immune-related miRNAs showed differential histone modifications in the promoter region. Our results indicated heightened inflammation in the susceptible line during early cytolytic MD, while resistant birds showed recuperative symptoms during early MD and epigenetic silencing during latent infection. Thus, although further elucidation of underlying mechanisms is necessary, this work provided the first definitive evidence of the epigenetic effects of MD.
  • Thumbnail Image
    Item
    A STUDY OF DIETARY PATTERNS IN THE MEXICAN-AMERICAN POPULATION AND THEIR ASSOCIATION WITH OBESITY
    (2006-08-17) Carrera Zamalloa, Patricia Margot; Mehta, Mira; Nutrition; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Ethnic groups have different dietary patterns based on their geographical locations and various cultural influences. We examined the dietary patterns of Mexican-Americans and their association with total and central obesity. We hypothesized that Mexican-American adults following a traditional diet would have a lower prevalence of obesity than those following a more typically American diet. Data from the NHANES 2001-2002 included 835 Mexican-American adults, aged 18 y and older. Dietary patterns were defined by cluster analysis of food group variables, expressed as percentage contribution to total energy intake. Obesity was assessed by body mass index (BMI, kg/m²) and central obesity by waist circumference. Surprisingly, we did not identify a "healthy pattern" group in this population, as has been generally observed in other ethnic groups. Contrary to our hypothesis, the traditional diet pattern was associated with higher values of BMI and waist circumference.
  • Thumbnail Image
    Item
    A Comparison Of Artificial Neural Networks And Statistical Regression With Biological Resources Applications
    (2006-08-07) Resop, Jonathan Patrick; Montas, Hubert J; Biological Resources Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Artificial neural networks (ANNs) have been increasingly used as a model for streamflow forecasting, time series prediction, and other applications. The high interest in ANNs comes from their ability to approximate complex nonlinear functions. However, the "black-box" nature of ANN models makes it difficult for researchers to design network structure or to physically interpret the variables involved. Recent investigations in ANN research have found connections linking ANNs and statistics-based regression modeling. By comparing the two modeling structures, new insight can be gained on the functionality of ANNs. This study investigates two primary relationships between ANN and statistical models: the potential equivalence between feed-forward neural networks (FNN) and multiple polynomial regression (MPR) models and the potential equivalence between recurrent neural networks (RNN) and auto-regressive moving average (ARMA) models. Equivalence is determined through both formal and empirical methods. The real-world phenomenon of streamflow forecasting is used to verify the equivalences found. Results indicate that both FNNs and RNNs can be designed to replicate many regression equations. It was also found that the optimal number of hidden nodes in an ANN is directly dependant on the order of the underlying physical equation being modeled. These simple relationships can be expanded to more complex models in future research.
  • Thumbnail Image
    Item
    FORAGING STRATEGIES, USE OF SPACE AND AGGRESSIVE BEHAVIOR OF DOMESTIC FOWL (Gallus gallus domesticus)
    (2004-05-03) Hoerl, Erin Natalie; Estevez, Inma; Animal Sciences
    Domestic fowl were tested in three experiments, all investigating the mediating effects of three group sizes of 5, 10, and 20 individuals, on behavioral responses under varying environmental conditions. The first experiment investigated social spacing and aggressive behavior in the presence or absence of cover panels. Smaller group sizes were more affected by cover panels than larger group sizes. In the second and third experiments patchy environments were used to test optimal foraging strategies. In the second experiment, smaller group sizes were more affected by patch locations than larger ones. In the third experiment birds were presented with patches varying in quality. Birds in all group sizes were able to immediately discern patch quality and preferred patches of higher quality. Despite generations of artificial selection pressure domestic fowl continue to forage optimally in patchy environments, and adopt flocking strategies predicted by behavioral ecology theory.