Severity scoring systems are frequently used in hospital intensive care units to assess patient wellness and mortality probability. Accurate mortality predictions are vital to provide appropriate and timely treatments to critical patients. However, commonly used severity scores have been found to make inaccurate mortality predictions. In this paper, we assess and compare four popular severity scores for both discrimination and calibration. We apply logistic regression, random forest, and neural network classification techniques in order to present new mortality prediction models, and compare their performance with pre-existing scores. We also compare the use of a basic set of predictor features that can be easily collected in the ICU environment with an expanded predictor set including laboratory diagnostics. Newly developed models improved on existing severity scores in terms of discrimination, with random forest providing the best results, but often demonstrated poor calibration. The expanded variable set did not improve model performance.

Employing over 10 million people, with an annual expenditure averaging over $1.2 trillion, the U.S. construction industry is massive in both scale and reach. Every day, thousands of construction builds are underway across the country, each with a unique set of costs contributing to its overall budget. For many, the cost of commodities, or materials used to build a project, is a huge variable that changes constantly and can negatively impact a project both financially and temporally. Being able to better understand and predict the way the prices these commodities change over time could lead to huge savings of time and money for anyone at the planning stages of a construction project. Statistical time series modeling is one effective way to increase this understanding and prediction abilities. This thesis models and forecasts the price indices of three construction commodities integral to the industry: lumber, iron/steel, and concrete. For each commodity, the month-to-month percentage changes of nearly 100 years of index values are modeled using Autoregressive Moving Average (ARMA) models, which are selected based on multiple model selection criteria. The models are then used to forecast an 8-month period directly following the end of the data series, which corresponds to the first 8 months of 2022. In doing so, it is found that the lumber model using ARMA methods makes forecasts indicative of what has happened in reality in the first half of 2022, where as the iron/steel and concrete models have much more difficulty in doing so.

Data analytics and machine learning have become vital applications in the world of sports, providing glimpses of statistically probable, mathematically formulated future states while helping teams and individual athletes make smarter decisions. The sport of tennis, featuring a relatively low number of moving parts, a bevy of readily available data and a global interest, is ripe for advanced forecasting analysis. In the following, the method of adaptive least squares is leveraged in conjunction with Kalman filtering to create time-variant match statistics for professional tennis players. Once the adaptive model is constructed, the sigmoid function is utilized to transform forecasted delta set values into forecasted probabilities of winning for any given matchup. The most successful model constructed from the web-scraped pool of data is continuously improving and correctly forecasts tennis match outcomes 63.54% of the time, exceeding the prediction accuracy if outcomes were chosen solely based on professional rankings.

It is common to use geostatistical techniques to analyze epidemiological data. However, we might gain further insight by viewing these types of data as a point pattern due to the spatial nature of the dataset which would allow us to use the spatial information of each point and apply point process techniques. Point process techniques are applied to the avian influenza virus data, and a summary statistic called the weighted cross J-function is proposed. The ordinary cross J-function is extended to a weighted version by incorporating weights to account for inhomogeneity because this dataset appears to exhibit non-constant intensity. Unlike the ordinary cross J-function, the weighted cross J-function takes into account the varying background rate of the point process by incorporating weights for each point in the point patterns. The advantage of the weighted cross J -function is that it is used to measure the interaction between events in two point processes and to detect clustering or inhibition between them, in order to recognize where spatial interaction appears most prevalent. We introduce the weighted cross J-function, discuss its properties, demonstrate it with simulations and show its application to the African Avian Flu dataset.

Point process modeling is used to predict the conditional intensity of wildfire ignitions in California from 2008 to 2012. Weather variables from the closest station record are used as model covariates. To reduce computation requirements of maximum likelihood estimation, the Stoyan-Grabarnik statistic is employed. Models of various combinations and number of variables are examined, as well as models for specifically naturally occurring fires caused by lightning and fires caused by humans.

AMP is a web page format that Google has developed for optimizing search results and mobile display. In this thesis, we will analyze the daily data for revenues, pageviews, and RPMs of a particular website that uses both AMP and regular web pages from 2016 to 2018. After data cleaning and transformation, we will utilize spectral analysis, regressions, multiplicative decomposition using LOESS, ARIMA modeling and other methods to analyze the time-series data, separate the seasonality and trend components, and build two different models and generate predictions. We will also analyze the specific impact of suspending AMP-enabled pages on the advertisement revenue of a website, and provide casual inference on the behavior of total revenue immediately after suspending AMP-enabled pages.

Basketball players are traditionally grouped into five distinct positions, but these designations

are quickly becoming outdated. We attempt to reclassify players into new groups

based on personal performance in the 2016-2017 NBA regular season. Two dimensionality

reduction techniques, t-Distributed Stochastic Neighbor Embedding (t-SNE) and principal

component analysis (PCA), were employed to reduce 18 classic metrics down to two dimensions

for visualization. k-means clustering discovered four groups of players with similar

playing styles. Player representation in each of the four clusters is similar across the 30

NBA teams, but better teams have players located further away from cluster centroids on

the scatterplot. The results indicate that strong teams have players whose success cannot be

attributed to fundamentals alone, meaning these players have advanced or intangible factors

that supplement their performance.

Earthquake occurrences are often described using a class of branching models called Epidemic-Type Aftershock Sequence (ETAS) models. The name derives from the fact that the model allows earthquakes to cause aftershocks, and then those aftershocks may induce subsequent aftershocks, and so on. Despite their value in seismology, such models have not previously been used in studying the incidence of invasive plant and animal species. Here, we apply a modified version of the space-time ETAS model to study the spread of an invasive species red banana trees (Musa velutina) in a Costa Rican rainforest. One challenge in this ecological application is that fitting the model requires the originations of the plants, which are not observed but may be estimated using filed data on the heights of the plants on a given date and their empirical growth rates. The formulation of the triggering density function, which describes the way events cause future occurrences of events is based on plots of inter-event times and distances for the red banana plants. We characterize the estimated spatial-temporal rate of spread of red banana plants using a space-time ETAS model. We then assess the triggering density more carefully using a non-parametric stochastic declustering method based on Marsan and Lengline (2008). When the algorithm is applied to the red banana data, the results indicate similar temporal and spatial structure, compared to previous estimates, as well as triggering of offspring running primarily to the northwest and the southeast from each parent. Non-parametric results are also used to obtain estimates of the most likely targets where immigration of red banana plants may be occurring.

Major League Baseball teams play considerably more games per season than the teams in any of the other major professional sports leagues in the United States, providing baseball fans with a large selection of games and frequently resulting in large fluctuations in attendance. In addition to studying the effect of the home and visiting teams' season win percentages on attendance, this paper examines various shorter-term metrics of performance to see which recent performance metric has the most significant impact on a home team’s game-to-game attendance numbers. The number of wins in a team’s previous 10 home games is shown to be statistically significant and the best recent performance metric to estimate attendance; however, the attendance prediction benefits are not practically much better than a simpler model that ignores such recent performance. This paper also discusses how the magnitude of the variation in a team’s attendance numbers can differ from team to team and from season to season and suggests using a team-season standardized attendance response variable to eliminate this difference in variance issue between low average / low variance, high average / high variance, and high average / low variance teams. The models with the team-season standardized response variable consistently outperform models with the traditional attendance response variable.

This paper aims to apply machine learning methods to analyze the biomarkers of symptoms in patients with schizophrenia. By reducing the dimension of brain patterns via random forest models and mapping brain patterns to symptoms of schizophrenia using multivariate regression models, we will explore the relationship between brain patterns and symptoms of schizophrenia and the association between different types of antipsychotic medication and brain patterns and symptoms.