DataFest Spring 2019 projects were grouped around five themes:
- Theme 1: Social Networks and Individual Characteristics in Multiplayer Games
- Overview of Multiplayer Games Dataset
- Multiplayer Game’s Solo Players
- Game Data and Social Capital
- Who is the One staying?
- Who is the Best Game Mentor?
- Theme 2: Data Science for Social Impact
- Theme 3: Our City
- Theme 4: Environmental Data Science
- Theme 5: Health and Well Being
Below is a summary of the final presentations of all the projects, which was held on April 26, 2019. For each project we include a description, the faculty advisors, and the students who contributed.
Theme 1: Social Networks and Individual Characteristics in Multiplayer Games
Overview of Multiplayer Games Dataset
Faculty advisor: Dmitri Williams, Associate Professor of Communication
– Online multiplayer games provide a wealth of data that can be used to study human behaviors. Professor Williams describes the kinds of questions that can be investigated with rich datasets of online game player actions, interactions, and targeted survey questions. Students in his group focused on a wide range of projects that use this data to study a range of human behaviors.
Multiplayer Game’s Solo Players
Students: Do Own (Donna) Kim
Faculty advisor: Dmitri Williams, Associate Professor of Communication
– Even when engaging in a multiplayer online game, some players play by themselves. Do Own is interested in investigating personality, motivation, and behavioral patterns of social network isolates.
Game Data and Social Capital
Students: Natalie Jonckheere & Calvin Liu
Faculty advisor: Dmitri Williams, Associate Professor of Communication
– Natalie and Calvin look at social capital from two angles: social capital as a predictor and social capital as an outcome. First, they want to observe whether people who display social capital exhibit certain characteristics or behavior. Second, they want to study if people who are interested in a specific topic will exhibit certain types of social capital.
Who is the One Staying?
Students: Qiyao (Joyce) Peng & Alejandro Marin
Faculty advisor: Dmitri Williams, Associate Professor of Communication Program
– Motivation plays a strong role as a moderator in the relationship between gamers’ in-game performances and enjoyment and churn, respectively. The research question for this project is ‘what is the relationship between players’ competitiveness and their level of enjoyment/churn?’
Who is the Best Game Mentor?
Student: Joo-Wha Hong
Faculty advisor: Dmitri Williams, Associate Professor of Communication
– This project explores the influence of the personality of game players who become mentors on mentoring outcomes using machine learning. The project will use survey data to analyze mentors’ extraversion and agreeableness as well as mentees’ game performance and churn rates.
Theme 2: Data Science for Social Impact
Foster care children: Administrative data and computational methods
Student: Eunhye Ahn
Faculty advisor: Emily Putnam-Hornstein, Associate Professor, School of Social Work
– This project is using population-based administrative data, including birth, medical, and education records to study child welfare services. The research topics include: a family-level analysis of first births and sibling re-reports in the foster care system; identifying mothers who gave subsequent birth after the termination of parental rights; modeling the child protective services system using Markov models; and predicting risks for aging youth.
Theme 3: Our City
Crosstown
Students: Giorgos Constantinou & John Cutone
Faculty advisors: Gabriel Kuhn, Professor of Journalism, Annenberg School of Communications; Seon Kim, Associate Director at USC’s Integrated Media Systems Center, Viterbi School of Engineering
– Giorgos and John are developing a machine learning system to automatically detect, prioritize, and alert journalists in the presence of abnormalities in crime data. Through this project, they want to assist journalists to identify interesting stories in data. The data is rich in features, and by using general feature engineering.
Theme 4: Environmental Data Science
A Tutorial on R and R Studio for environmental sciences curriculum
Student: Huy Nghiem
Faculty advisors: Jill Sohm, Assistant Professor (Teaching) of Environmental Studies, Dornsife College of Lettters, Arts, and Sciences; Deborah Khider, Data Scientist, Information Sciences Institute, Viterbi School of Engineering.
– This project created an electronic notebook using R and R studio for environmental sciences curriculum. The notebook will be used for undergraduate students to teach advanced statistical analysis about population health in Fall 2019.
Analyzing paleoclimate data
Student: Han Wu
Faculty advisor: Deborah Khider, Data Scientist, Information Sciences Institute, Viterbi School of Engineering
– This project focused on a causality analysis of paleoclimate time series using Pyleoclim, which is a Python package geared towards the analysis and visualization of paleoclimate data. Future work includes exploring and testing additional algorithms for time series analysis.
Theme 5: Health and Well Being
Lighting Control in Buildings for Visual Comfort
Student: Lingkai Cen
Faculty advisor: Joon-Ho Choi, Assistant Professor of Building Science, School of Architecture
– Lingkai collected data in a controlled setting and studied lighting control in buildings for visual comfort. For data preprocessing, he used Excel and MathWorks; for data analysis, he used Python and Scikit learn.
Enhancing Thermal Control in Buildings
Student: Mengqi Jia
Faculty advisor: Joon-Ho Choi, Assistant Professor of Building Science, School of Architecture
– Mengqi presented her study about thermal control using Excel, Minitab, and Python for data analysis. She collected data in a lab setting and tried two models, a group model and an individual model. She found that a group model, which consisted of 20 subjects, did not work well, while individual models gave better results.
Invited Projects
Behavioral Context Recognition (Invited GRIDS project)
Student: Ian Myoungsu Choi
– This project studied behavioral patterns by analyzing data from personal senses collected from 60 subjects. This data can be used to predict activities and infer people’s lifestyle and habits.