Attempts to determine the factors that influence an individual’s decision to visit a lake are of increasing importance as more detailed statistics are collected for a large number of lakes across the United States. The present research seeks to use the recently compiled LAGOS dataset to create a functional model lake visitation, with a special attention paid to water clarity and the number of lakes nearby a given lake. This modeling found that both clarity and nearby lakes were significant predictors in determining whether a given lake is visited, as well as being significant factors in estimating total lake visitation. In addition, three models of water clarity were constructed with a focus placed on lake chemistry. The first focused directly on modeling lake clarity, using a linear mixed effects model to account for multiple measurements taken at a single lake. This model found significant negative effects for chlorophyll and phosphorus, as well as an effect for region in the US. The second model sought to determine factors that can be used to predict eutrophic status in 5 years for a given lake. This model found significant negative effects for initial water clarity, and positive effects for chlorophyll change and phosphorus change. The last model attempts to determine what features of a lake can serve as predictors as to whether or not the lake is measured for clarity. This model found that region of the United States and population surrounding the lake were significant predictors.

Research has demonstrated that group identification creates a desire to distinguish the in-group from the out-group. A well-known mechanism of group assignment, the minimal groups paradigm, has been used to show that group differentiation occurs even when no actual distinction exists between the in-group and out-group. However, little research has been conducted looking at whether individuals perceive members of their own in-group as being characteristically different than members of the out-group, using a minimal groups assignment. The present research evaluated whether individuals differentiate in-group and out-group members on personality and character traits, even when all real differences have been removed using minimal groups. This study found that members of the out-group were rated more positively on average than members of the in-group. Additionally, members of both groups were rated more positively on socially favorable traits than less favorable or arbitrary traits. These findings suggest that minimal groupings may not be sufficient to create an association between an assigned group member and his or her in-group. In this case, heuristics such as out-group homogeneity bias and trait favorability may become more salient in the group judgment process.

As statisticians, we are fundamentally interested in the relationships between variables. Per- haps the most common tool for examining these relationships is the scatterplot. Because the human brain has evolved to detect complex patterns, a viewer can use scatterplots to characterize a wide variety of relationships between variables. But as the number of variables in a dataset grows, it becomes infeasible for a single viewer to consider all possible scatterplots. Furthermore, this tendency toward pattern detection can lead to the identification of relationships that do not exist in reality. In this paper, we will discuss how a computer can be trained using statistical learning methods to detect patterns in scatterplots more efficiently and effectively than humans. Our research leads to the development of an automated procedure to detect relationships between variables in large data sets and opens the door to more powerful data analysis.

New York City is home to one of the largest public transportation networks in the world, the foundation of which is the arsenal of yellow and green taxi cabs that line nearly every street corner. However, with the advent of ride share apps, cab companies are struggling to remain competitive in this diverse transit system. This research analyzes over 1.5 billion taxi trips in New York City from January 2009 to June 2016 in order to predict areas of the city and times of the day in which there is a mismatch between the number of available cabs and the number of awaiting passengers. Then, we use these predictions to devise an optimal dispatch algorithm that best allocates empty taxi cabs to potential customers in order to optimize the efficiency of the taxi fleet as a whole.