Data Science Commons^™

How Graduate Student Fellows Enhance What A Center For Digital Scholarship Does, Ben B. Chiewphasa

Transforming Libraries for Graduate Students

Multiple disciplines are increasingly embracing data science and digital scholarship. However, insufficient training for digital and computational methodologies within subject/departmental silos means that these needs often get overlooked. Opportunities for learning how to teach technical concepts (i.e., how to handle troubleshooting, live participatory coding, etc.) are also rare or non-existent via departmental offerings. To respond to these needs, the Navari Family Center for Digital Scholarship launched its Pedagogy Fellowship Program in Fall 2021 where Notre Dame PhD students/candidates build their instructional expertise and experience related to digital scholarship with an added bonus of enhancing their competitiveness on the job market. …

Directional Pairwise Class Confusion Bias And Its Mitigation, Sudhashree Sayenju, Ramazan Aygun Phd, Jonathan Boardman, Duleep Prasanna Rathgamage Don, Yifan Zhang Phd, Bill Franks, Sereres Johnston Phd, George Lee, Dan Sullivan, Girish Modgil Phd

Published and Grey Literature from PhD Candidates

Recent advances in Natural Language Processing have led to powerful and sophisticated models like BERT (Bidirectional Encoder Representations from Transformers) that have bias. These models are mostly trained on text corpora that deviate in important ways from the text encountered by a chatbot in a problem-specific context. While a lot of research in the past has focused on measuring and mitigating bias with respect to protected attributes (stereotyping like gender, race, ethnicity, etc.), there is lack of research in model bias with respect to classification labels. We investigate whether a classification model hugely favors one class with respect to another. …

Explainabilityaudit: An Automated Evaluation Of Local Explainability In Rooftop Image Classification, Duleep Rathgamage Don, Jonathan Boardman, Sudhashree Sayenju, Ramazan Aygun, Yifan Zhang, Bill Franks, Sereres Johnston, George Lee, Dan Sullivan, Girish Modgil

Published and Grey Literature from PhD Candidates

Explainable Artificial Intelligence (XAI) is a key concept in building trustworthy machine learning models. Local explainability methods seek to provide explanations for individual predictions. Usually, humans must check these explanations manually. When large numbers of predictions are being made, this approach does not scale. We address this deficiency for a rooftop classification problem specifically with ExplainabilityAudit, a method that automatically evaluates explanations generated by a local explainability toolkit and identifies rooftop images that require further auditing by a human expert. The proposed method utilizes explanations generated by the Local Interpretable Model-Agnostic Explanations (LIME) framework as the most important superpixels of …

Integrated Gradients Is A Nonlinear Generalization Of The Industry Standard Approach To Variable Attribution For Credit Risk Models, Jonathan Boardman, Md Shafiul Alam, Xiao Huang, Ying Xie

Published and Grey Literature from PhD Candidates

In modern society, epistemic uncertainty limits trust in financial relationships, necessitating transparency and accountability mechanisms for both consumers and lenders. One upshot is that credit risk assessments must be explainable to the consumer. In the United States regulatory milieu, this entails both the identification of key factors in a decision and the provision of consistent actions that would improve standing. The traditionally accepted approach to explainable credit risk modeling involves generating scores with Generalized Linear Models (GLMs) - usually logistic regression, calculating the contribution of each predictor to the total points lost from the theoretical maximum, and generating reason codes …

A Distance-Based Clustering Framework For Categorical Time Series: A Case Study In Episodes Of Care Healthcare Delivery System, Lauren Staples

Doctor of Data Science and Analytics Dissertations

Understanding how compensation structures influence overall healthcare costs is a central issue in health economics. Episodes of Care (EoC) is a compensation structure that bundles payments for healthcare interventions that belong to a well-defined health event. Since the variation of clinical pathways can drive the cost of healthcare, this research uses sequences of medical billing codes in Perinatal Episodes of Care claims data to study the extent of that variation by equating it to the number of reproducible clusters found. This research proposes a methodological framework to detect reproducible clusters in an unsupervised problem where the true number of clusters …

Anti-Vaxxers: Parents Fighting Science, Katie West

Symposium of Student Scholars

Immunizing children helps protect the health of our community, especially those people who cannot be immunized. Yet, since 1996 after a study was released that linked autism to vaccinations, there has been a trend of parents refusing to vaccinate their children. What are the demographics of the parents who believe their children are better off without vaccines? By knowing where these parents live and what decisions they make for their children’s education, counties and medical professionals can provide education and address their concerns.

My research involves data on 116,141 kindergarten classes from 2000-2015 in California. The two vaccine exemption options …

Opioid Abuse: Are Doctors Creating The Problem?, Nguyen Tran

Symposium of Student Scholars

Opioid abuse and overdose are serious health problems in the United States. Current research has concentrated on the treatment and prevention of opioid abuse. Using data from the Controlled Substance Utilization Review and Evaluation System (CURES) for California zip codes, my research focuses on the causes of opioid overdose by considering the relationships between the following variables within each zip code: population size, average number of prescriptions per doctor, percentage of people who receive opioid prescriptions, percentage of people receiving the same prescription drug from 3 or more doctors, average number of opioid pills per prescription and number of people …

Market Research: How To Keep And Gain Customers, Chris Mccall

Symposium of Student Scholars

Customer-centered market research is essential to the creation and management of successful marketing campaigns. A company that understands their customers will be able to provide those customers with products and services that fit their needs better than the competition, and ultimately increase profits. My research focuses on a database containing customer information for a telecommunications company called Telco. Within this research, I will focus on a number of customer attributes including demographics, services provided, payment methods, contract lengths, monthly charges, and tenure with the company. Considering how these attributes relate to one another will give me a better understanding of …

Food Deserts: Hungry For Answers, Lawren Cumberbatch

Symposium of Student Scholars

In 2010, the United States Department of Agriculture (USDA) reported that 23.5 million people in the United States live in food deserts. As defined by the USDA, a “food desert” is a neighborhood that lacks healthy food sources. This can be measured by distance to a store, number of stores in an area, individual-level resources such as family income or vehicle availability, and neighborhood-level resources such as availability of public transportation. Past research provides evidence that food deserts are especially likely to occur in communities heavily populated by minorities. As a Black Indian pre-med student aiming to join the world …

Determining Malignancy: Can Mammogram Results Help Predict The Diagnosis Of Breast Tumors?, Taylor Behrens

Symposium of Student Scholars

Even with advancements in treatment and preventative care, breast cancer remains an epidemic claiming more than 40,000 American male and female lives each year. The mammogram dataset that I am analyzing was initially complied in the early 1990s by a team from the University of Wisconsin - Madison. Past research diagnoses breast cancer from fine-needle aspirates. My research focuses on predicting whether we can determine breast cancer diagnoses without the use of invasive procedures and, in particular, whether we can predict breast cancer based on mammogram data. Do measures of gray-scale texture, radius, concavity, perimeter, compactness, area, and smoothness of …

Accidental Overdoses: Insights To Aid In Prevention, Annabel Nganga

Symposium of Student Scholars

Having lost a friend six years ago to an accidental cocaine overdose, I am very passionate about spreading awareness of accidental drug overdoses that have affected thousands of families countrywide. According to past research, deaths resulting from opiates specifically have been on the rise, and a significant number of deaths in the United States for those below fifty years are caused by drug overdoses. Data exists indicating which states have more overdoses. The data set I will be using includes variables on race, sex, age, drug with which person overdosed, location of the overdose, ultimate cause of death and year …

Death By Police: When “Protecting And Serving” Goes Wrong, Hesper Mallis

Symposium of Student Scholars

The recent cases of law enforcement using lethal force in the United States have gained massive public attention. My dataset is from the Mapping Police Violence website. The website’s focus was to create a heat map to display where police killings occurred most frequently. The website has a dataset with information on 7,664 deaths of suspects. The variables in the dataset include age, sex and race of the suspect; geographic location; alleged threat level; alleged weapon; cause of death; and criminal charges against the officer. In addition, the variables include whether the individual had a mental illness, was armed or …

Are There Predictors Of A Running Back’S Success?, Joshua Price

Symposium of Student Scholars

People who analyze football have concentrated in the past on a running back’s 40-yard dash, shuffle, broad jump, vertical jump, and bench press measures. My research will test if the following variables can predict a running back’s success in the NFL: height, weight, conference, offensive line ranking for their team, the running back’s total yards for the season, their average yards for each attempt, the number of times the running back has entered the end zone for a touchdown that season, the running back’s time average time behind the line of scrimmage (TLOS), the percentage of times the running back …

Sources And Aftermaths Of Pipeline Related Leaks And Spills, Justin Smith

Symposium of Student Scholars

The escape of oil and other hazardous materials have been shown to pollute and destroy ecosystems. As an aspiring chemist, I am adamant about the secure handling and transportation of oil and other hazardous materials. In the past, researchers have concentrated on oil’s high viscosity. Oil’s high viscosity physically smothers wildlife, affecting their ability to continue critical functions such as respiration, feeding, and thermoregulation. My research focuses on the source of these oil spills, as well as natural gas leaks, for the purpose of risk assessment. In addition, I compare recovery efforts based on the cause of the leak/spill, the …

On The Front Lines Of Fire: How Do We Save Their Lives?, Cathrine Jatta

Symposium of Student Scholars

The National Institute for Occupational Safety and Health (NIOSH) reports that the United States depends on about 1.1 million firefighters to protect its citizens and property from fire. NIOSH adds that approximately 336,000 are career firefighters; 812,000 are volunteers; and 80 to 100 die in the line of duty each year. NIOSH investigates each fatality individually for the cause and prevention. In contrast, my research will look at a complete dataset of 2005 firefighter fatalities and see if any of the following variables may predict firefighter death: age, cause of death, property type, type of duty (e.g. on-duty, training), and …

Cervical Cancer: Are There Ways To Reduce The Risks?, Madelyn Dorn

Symposium of Student Scholars

History has shown us that when caught early, cervical cancer is curable. Past research has found that the sexually transmitted diseases (STDs), herpes and human papillomavirus (HPV), have been associated with cervical cancer. In contrast, my dataset on 859 women has many more STDs and lifestyle choices compiled on 36 variables. The diagnoses in the dataset are many: cervical condylomatosis, vaginal condylomatosis, vulvo-perineral condylomatosis, syphilis, pelvic inflammatory disease, genital herpes, molluscum contagiosum, acquired immune deficiency syndrome (AIDS), human immunodeficiency virus (HIV), hepatitis B, HPV, and cervical cancer. In addition to the demographic variable on age, there are many lifestyle choice …

Marijuana Arrests In Toronto Canada: A Look Into The Canadian Criminal Justice System, Steven Tully

Symposium of Student Scholars

Marijuana related drug offenses made up fifty-eight percent of all Controlled Drugs and Substances Act offenses in Canada in 2016. On October 17, 2018, Canada legalized marijuana. As part of the efforts to legalize marijuana, descriptive statistics of single variables, like the age of the arrestees and the number of people arrested per year, were reported by the Toronto Star newspaper. The dataset analyzed in this research predates the legalization of marijuana and was collected from 1997 to 2002 on 5,226 individuals arrested in Toronto, Canada for simple possession of small quantities of marijuana. When an offender was arrested for …

Who Is Next? Evaluating Factors That May Contribute To Heart Failure, Davon Broadwater

Symposium of Student Scholars

Cardiovascular diseases are the number one causes of death globally, and for African Americans those risks are even higher. As an African American university student studying Biology, I am passionate about researching the diseases that affect my race. Current research states that behavioral factors such as obesity, tobacco use, unhealthy diet, and harmful use of alcohol should be avoided. I have chosen to research predictors of what helps patients survive if they already have heart failure. Heart failure develops gradually, where the heart becomes weaker over time and has trouble pumping blood to nourish the cells in the body. Data …

Eradicating Zebra Mussels: What Works?, Elijah Davies

Symposium of Student Scholars

The invasion of U.S lakes and rivers by the invasive species of zebra mussels called Dreissena polymorpha has caused catastrophic harm to the local ecosystem by reproducing and outcompeting native mussel species as well as harm to pipes leading into water sources by binding to surfaces and reproducing to the point that the mussels clog pipes. In addition, recreation areas must be closed due to the sharp shells making areas unusable. In the past, research has focused on individual molluscicides and their eradication of zebra mussels, as well as their effect on native flora and fauna. My research will contrast …

Bias In Police Shootings: Is It Just An Opinion?, Phuong Ho

Symposium of Student Scholars

The claims of racism have drawn public attention toward police brutality and its impact on minorities. Is this just an opinion or is there any statistical evidence? Recent studies from The Atlantic have investigated the average age and ethnicity of victims from police killings in 2015-2016. As an Asian-American, I am motivated to examine the issue of police killings among races and other demographics to find any bias that is present. Using the dataset of 2,204 victims of police killings (2015-2016) collected by The Guardian, I will examine the following variables for bias: age, cause of death, armed/unarmed, race/ethnicity, and …

Do Environmental Toxins Predict Violent Crimes?, Tyler Stahl

Symposium of Student Scholars

Do chemical pollutants that persistent in the environment and bioaccumulate in the body affect human health and behavior? Could these Persistent, Bioaccumulative, and Toxic (PBT) chemicals play a role in the cause of violent crimes due to deterioration of mental and cognitive functions? In the past, Mercury, a PBT chemical, has been shown in salmon to be associated with aggression. Could similar aggression occur in humans exposed to mercury through a toxic spill? Two sources of data are utilized in this analysis. The Environmental Protection Agency’s (EPA) Annual Toxic Release Inventory publishes data on toxic releases into the environment and …

Using Big Data Analytics To Optimize Practical Large Databases, Po-Chun Lu

Master of Science in Computer Science Theses

Big data analytics is gaining popularity for enterprises in optimizing their business processes ranging from retailers, supply chains, to online shopping stores. Existing practical raw data are far from usable to achieve the goal. Therefore, a good data pre-processing approach is required and is a key step to success. We propose to research on the effectiveness of data pre-processing and the business process based on a real world database. Our methodology involves natural language processing. Our key goal is to study appropriate methods with big data analysis techniques that can handle errors, ambiguity, and repeated descriptions caused by human languages. …

Reporting Of Eating Disorder Deaths, Katherine Mobley, Amy Hord

Symposium of Student Scholars

Those affected by eating disorders experience disturbances in eating behaviors which are often related to underlying psychiatric disorders such as anxiety, depression, or obsessive-compulsive disorder (Parekh, 2017, Drieberg et al., 1998 p.53). The duplicitous nature of the disorder makes it difficult to diagnose, and the tole it takes on an individual’s physical health makes its mortality rate the second highest among psychiatric disorders (Guinhut et al., 2021 p.130). Even if the correct education and resources are accessible to certain individuals, negative stigmatization about the disorder can make sufferers unlikely to seek help (Becker et al., 2010). Findings from analysis of …

The Social Market Economy As A Formula For Peace, Prosperity, And Sustainability, Almuth D. Merkel

Doctor of International Conflict Management Dissertations

The social market economy was developed in Germany during the interwar period amidst political and economic turmoil. With clear demarcation lines differentiating it from socialism and laissez-faire capitalism, the social market economy became a formula for peace and prosperity for post WWII Germany. Since then, the success of the social market economy has inspired many other countries to adopt its principles. Drawing on evidence from economic history and the history of economic thought, this thesis first reviews the evolution of the fundamental principles that form the foundation of social-market economic thought. Blending the micro-economic utility maximization framework with traditional growth …

Federated Learning For Secure Sensor Cloud, Viraaji Mothukuri

Master of Science in Software Engineering Theses

Intelligent sensing solutions bridge the gap between the physical world and the cyber world by digitizing the sensor data collected from sensor devices. Sensor cloud networks provide resources to physical and virtual sensing devices and enable uninterrupted intelligent solutions to end-users. Thanks to advancements in machine learning algorithms and big data, the automation of mundane tasks with artificial intelligence is becoming a more reliable smart option. However, existing approaches based on centralized Machine Learning (ML) on sensor cloud networks fail to ensure data privacy. Moreover, centralized ML works with the pre-requisite to have the entire training dataset from end-devices transferred …

Fine-Grained Sentiment Analysis For Customer Review, Bing Han, Meng Han, Jing (Selena) He

Master of Science in Computer Science Theses

Natural Language Processing (NLP) is one of the most attractive technologies in many applications in real-life. Sentiment analysis, which has devoted to know others' think or feel about an experience or an item and hence take an action, is one of the most developed area in both academia and industry. Among sentiment analysis, fine-grained aspect sentiment analysis attempts to analyze emotional attitude categorized into different aspects or features of an(a) experience/service/product. Although aspect level sentiment analysis could provide more useful information, the proposed models' performance were relative poor compared with document-level or sentence-level sentiment analysis due to the lack of …

Profile Modeling In Hierarchical Deep Architecture By Mutual Support, Honglai Peng, Meng Han, Jing (Selena) He

Master of Science in Computer Science Theses

Despite significant advances in the field of face analysis over last decade, the current studies are still limited to specific face computation tasks using deep learning approaches. In this paper, we propose an end-to-end hierarchical deep learning structure, called Multi-Features Convolutional Neural Networks (MFCNN), which can comprehensively implement face analysis including age, gender, race and emotion. Moreover, we take the advantages of the mutual support among different facial features from individual tasks to improve the performance of our model. We also contribute one all-labeling dataset called Multiple Facial Features Computation (MFFC) based on Apparent-age-V2 dataset. Firstly, we train four different …

Improving The Maximum Power Point Tracking Efficiency Of Photovoltaic Arrays Via Machine Learning And Deep Learning, Sumedha Inamdar

Master of Science in Computer Science Theses

Under partial shading conditions, photovoltaic (PV) modules in a solar array experience varying irradiance. A Global Maximum (GM) and multiple Local Maximums (LMs) can originate on the Power-Voltage (P-V) curve under nonuniform irradiance conditions. There are many maximum power point tracking (MPPT) algorithms developed to detect the true maximum power point (MPP) of a PV array. However, in the real-world environment, limited samples of power-voltage (P-V) data might be available to quickly and accurately predict the position of the global maximum point. Since the change of environmental conditions are dynamic, limited time is available to locate the global peak. Machine …

Classifying Imbalanced Financial Fraud Data Utilizing Enhanced Random Forest Algorithm, Charles Gardner

Master of Science in Computer Science Theses

Imbalanced datasets have been a unique challenge for machine learning, requiring specialized approaches to correctly classify the minority class. Financial fraud detection involves using highly imbalanced datasets with a class imbalance of up to .01% frauds to 99.99% regular transactions. It is essential to identify all frauds in financial fraud detection, even if some classifications' precision is low. I developed a random forest assembly that separates fraudulent transactions into tiers of precision. With this approach, 96% of fraudulent transactions are identified, showing an 8% increase in recall when compared to standard approaches. 59% of fraud classifications' precision increases by 10% …

Data Mining And Image Classification Using Genetic Programming, Mahsa Shokri Varniab

Master of Science in Computer Science Theses

Genetic programming (GP), a capable machine learning and search method, motivated by Darwinian-evolution, is an evolutionary learning algorithm which automatically evolves computer programs in the form of trees to solve problems. This thesis studies the application of GP for data mining and image processing. Knowledge discovery and data mining have been widely used in business, healthcare, and scientific fields. In data mining, classification is supervised learning that identifies new patterns and maps the data to predefined targets. A GP based classifier is developed in order to perform these mappings. GP has been investigated in a series of studies to classify …