In this study I analyzed the pattern of settlement for known Medio period (A.D. 1200–1450) sites in the Casas Grandes region of Chihuahua, Mexico. Locational data acquired from survey projects in the Casas Grandes region were evaluated within a Geographic Information Systems (GIS) framework to reveal patterns in settlement and site distribution. Environmental and cultural variables, including aspect, cost distance to nearest ballcourt, ecoregion, elevation, local relief, cost distance to nearest oven, cost distance to Paquimé, slope, soil, terrain texture, topographic position index, cost distance to nearest trincheras, vegetation, vegetation variety to 100 meters, vegetation variety to 500 meters, cost distance to nearest intermittent lake, cost distance to nearest intermittent stream, cost distance to nearest perennial lake, and cost distance to nearest perennial stream were calculated for each site in this region. It was expected that the relationships of correspondence between known sites and these variables would provide a quantitative framework that could be used to model the locational probability of unknown sites in the region. Through the use of GIS and statistical analyses, the results of this study were used to produce an archaeological site sensitivity map for this region of northern Mexico.

In this thesis, we explore a multi-indexed logistic regression (MILR) model, with particular emphasis given to its application to time series. MILR includes simple logistic regression (SLR) as a special case, and the hope is that it will in some instances also produce significantly better results. To motivate the development of MILR, we consider its application to the analysis of both simulated sine wave data and stock data. We looked at well-studied SLR and its application in the analysis of time series data. Using a more sophisticated representation of sequential data, we then detail the implementation of MILR. We compare their performance using forecast accuracy and an area under the curve score via simulated sine waves with various intensities of Gaussian noise and Standard & Poors 500 historical data. Overall, that MILR outperforms SLR is validated on both realistic and simulated data. Finally, some possible future directions of research are discussed.

My Masters research aims to deepen our understanding of the behaviour of robust methods in logistic regression. Logistic regression is a special case of Generalized Linear Modelling (GLM), which is a powerful and popular technique for modelling a large variety of data. Robust methods are useful in reducing the effect of outlying values in the response variable on parameter estimates. A literature survey shows that we are still at the beginning of being able to detect extreme observations in logistic regression analyses, to apply robust methods in logistic regression and to present informatively the results of logistic regression analyses. In Chapter 1 I have made a basic introduction to logistic regression, with an example, and to robust methods in general. In Chapters 2 through 4 of the thesis I have described traditional methods and some relatively new methods for presenting results of logistic regression using powerful visualization techniques as well as the concepts of outliers in binomial data. I have used different published data sets for illustration, such as the Prostate Cancer data set, the Damaged Carrots data set and the Recumbent Cow data set. In Chapter 4 I summarize and report on the modem concepts of graphical methods, such as central dimension reduction, and the use of graphics as pioneered by Cook and Weisberg (1999). In Section 4.6 I have then extended the work of Cook and Weisberg to robust logistic regression. In Chapter 5 I have described simulation studies to investigate the effects of outlying observations on logistic regression (robust and non-robust). In Section 5.2 I have come to the conclusion that, in the case of classical or robust multiple logistic regression with no outliers, robust methods do not necessarily provide more reasonable estimates of the parameters for the data that contain no st~ong outliers. In Section 5.4 I have looked into the cases where outliers are present and have come to the conclusion that either the breakdown method or a sensitivity analysis provides reasonable parameter estimates in that situation. Finally, I have identified areas for further study.

<p>In this study I analyze the Swedish parental benefit system focusing on a certain component, the basic level, and its effects on childbearing. The basic level consists of a fixed monetary amount given to parents. The compensation level was raised three times between 2002 and 2004, which would, according to economic theories, imply increased incentives for having children. I use longitudinal data of women in fertile age between 1993 and 2005. Applying a difference-in-differences model on comparison groups based on region of birth I conclude that the policy changes probably did not cause increased probabilities of having children.</p>

Maltreatment affected an estimated 794,000 children in the 50 States, the District of Columbia, and Puerto Rico in 2007 (United States Department of Health and Human Services [USDHHS], 2009). The purpose of this study was to examine the risk factors of young maternal age, parents' marital status, multiple birth, preterm birth, birth defects/disability, low economic status, and parental substance abuse related to suspected maltreatment of children 3 years of age or younger from the prospective of pediatric nurse practitioners (PNPs). A cross-sectional survey design, using the Tailored Design Method, was used in this study. A convenience sample consisting of the National Association of Pediatric Nurse Practitioners (NAPNAP) email registry was used for this study with a response rate of 11%. The respondents represented all regions of the United States. Seventy-nine percent of the PNP's (n=363) who completed the survey had suspected child abuse or neglect within the last year in a child three years of age or younger compared to 21% of PNPs (n=96) who did not suspect child maltreatment within the past year. The prevalence of suspected child maltreatment in the study population was 2.35%. According to the model examining child risk factors and abuse, the log of the odds of a child being abused was negatively related to preterm birth (p = .036) and birth defects/disability (p = .001). Multiple birth was positively related but not significant (p = .359). There were no statistically significant child risk factors found in the logistical regression for neglect (preterm birth, p = .180; multiple births, p = .938; birth defects/disabilities, p = .234). When examining the abuse and neglect groups together, the log of the odds of a child being abused and neglected was negatively related to birth defects/disabilities (p = .030). Preterm birth (p = .364) and multiple birth (p = .298) were positively related to the abuse and neglect group but were not significant. According to the model examining parental risk factors and abuse, the log of the odds of a child being abused due to a parent characteristic was negatively related to low economic status, with the proxy being WIC eligibility (p = .001) and a history of substance abuse (p = .031). The regression for abuse indicated a positive, yet insignificant, relationship with young maternal age (p = .129) and single marital status (p = .816). The logistic regression for neglect indicated a positive significant relationship with a substance abuse history (p = .012). The regression for neglect indicated positive but insignificant relationships for young maternal age (p = .693), marital status (p = .343), and WIC eligibility (p = .106). There were no statistically significant parental risk factors found in the logistical regression for abuse and neglect together (young maternal age, p = .263; marital status, p = .523; WIC eligibility, p = .131; substance abuse, p = .985). Findings indicated that child maltreatment is suspected by PNPs in primary care settings, and that PNPs recognize signs and symptoms of abuse and neglect.

In this study I analyze the Swedish parental benefit system focusing on a certain component, the basic level, and its effects on childbearing. The basic level consists of a fixed monetary amount given to parents. The compensation level was raised three times between 2002 and 2004, which would, according to economic theories, imply increased incentives for having children. I use longitudinal data of women in fertile age between 1993 and 2005. Applying a difference-in-differences model on comparison groups based on region of birth I conclude that the policy changes probably did not cause increased probabilities of having children.

This research investigated the association of street network connectivity differences across travel modes with travel behaviour – mode choice, distance traveled and number of trips. To date research on travel behaviour relationships with urban form has not developed empirical evidence on street designs as distinct networks for walking and driving. A street network having greater connectivity and continuity for the pedestrian mode of travel vis-à-vis the vehicular network, like the Fused Grid, will likely encourage more walking. This hypothesis was investigated using a quasi-experimental approach within a rational utility behavioural framework. Local travel behaviour is theorized to be affected by desire to access goods and services (broadly termed, ‘activities’) in the community where people live. Using inferential statistics, the research tested for relationships between measured street patterns and self-reported local travel by King County, WA households. The main variables were ratios (walking : driving) of network connectivity and density, in the vicinity of travel survey households. Demographics and household characteristics, as well as other behaviourally influential urban form factors (residential density, proximity of destinations, etc.), were included in regression models, allowing control for confounding factors. Findings suggest that street networks with connectivity that provides better routing for one mode of transportation over others encourage more travel by the favored mode. The regression model demonstrated that a change from a pure small-block grid to a modified grid (i.e. Fused Grid) can result in an 11.3% increase in odds of a home-based trip being walked. The modified street pattern like a Fused Grid is also associated with a 25.9% increase, over street patterns with equivalent route directness for walking and driving, in the odds a person will meet recommended levels of physical activity. Finally, the Fused Grid’s 10% increase in relative connectivity for pedestrians is associated with a 23% decrease in local vehicle travel distance (VMT), and its improved continuity is associated with increased walking trips and distance. Conclusions: Other factors being equal, residential street networks with either more direct routing for pedestrians or more pedestrian facilities relative to vehicular network are associated with improved odds of walking and reduced odds of driving.

In recent years, changes in government policies, supply and demand fundamentals and price patterns in the cotton market have led to several shifts in how producers market their cotton. This thesis examined producer cash marketing choices, including direct and indirect hedging, in four different periods since 2001. Special emphasis was placed on the 2010 season - a season characterized by historically high prices and volatility. Producer marketing behavior was modeled as a discrete choice between four different cash market outlets: forward contracting with a merchant, post-harvest cash contracting, contracting with a merchant pool and contracting with a cooperative pool. Hedging was characterized as a tool that was used in conjunction with one of the four discrete choices. This thesis employed multinomial logit estimation to determine the influence of factors on producers' choice of primary cash marketing decisions. Data were collected from a mail survey of the population of cotton growers in Texas, Oklahoma and Kansas. The most important determinants of cotton cash marketing choices were 1) prior participation in cooperative pools, beliefs about the value of pre-harvest pricing, beliefs about the performance of merchant pools, willingness to accept lower prices to reduce risk, and several socio-economic variables.

No description available.

The objective of this thesis is to evaluate different models and methods for website classification. The websites are classified based on their functionality, in this case specifically whether they are forums, news sites or blogs. The analysis aims at solving a search engine problem, which means that it is interesting to know from which categories in a information search the results come. The data consists of two datasets, extracted from the web in January and April 2013. Together these data sets consist of approximately 40.000 observations, with each observation being the extracted text from the website. Approximately 7.000 new word variables were subsequently created from this text, as were variables based on Latent Dirichlet Allocation. One variable (the number of links) was created using the HTML-code for the web site. These data sets are used both in multinomial logistic regression with Lasso regularization, and to create a Naive Bayes classifier. The best classifier for the data material studied was achieved when using Lasso for all variables with multinomial logistic regression to reduce the number of variables. The  accuracy of this model is 99.70 %. When time dependency of the models is considered, using the first data to make the model and the second data for testing, the accuracy, however, is only 90.74 %. This indicates that the data is time dependent and that websites topics change over time.