Rishi Adi, Pingyao Liu, Anondo Deepro Chowdhury, Prajwal Dacharla, Anna Khodakovskaia, Jennifer Luo, Jason Ma, and Adrian Shin
For the sixth year, the San Francisco Bay Area Chapter of the American Statistical Association held its summer program to help K–12 students understand statistics and its applications. Over several months, six student-mentor groups worked together to analyze data sets covering topics from sports to health science. The mentor volunteers helped drive the smooth progress of all the projects, culminating in a final virtual presentation.
The program began with a kick-off meeting in June. The students were assigned mentors with expertise in the analysis they wanted to conduct and started to break ground on their journeys. Some students chose to work with just their mentor, while others collaborated with a fellow student. After discussion, students picked topics that spoke the most to them and began finding publicly available data sets for training, testing, and evaluating the models they created. Over three months, the students and mentors met to share updates on the progress of their projects and work through challenges, ending with a final presentation.
Eleventh-grader Jason Ma and twelfth-grader Rishi Adi explored the effects of seasons and movie ratings on box office performance. Working with Cristina Tortora, a professor at San Jose State University, and junior Anwen Huang of the University of California, Berkeley, Ma and Adi analyzed two data sets about box office performance. They wrote multiple programs in statistical programming language R to analyze and draw conclusions from the data sets on how certain factors relate to whether a movie performs well at the box office. Both Ma and Adi used various statistical methods, such as analyses of variance and t-tests, to arrive at their conclusion. They determined movies typically do better at the box office if released in spring and summer, despite more films being released in theaters during autumn and winter.
Ma said, “Something that I found while performing the aforementioned analysis is the importance of computer organization skills, such as creating a meaningful file directory structure and having some file naming conventions. I am very grateful for the help that I have received from the SFASA mentors, especially Dr. Cristina Tortora, and volunteers throughout the summer.”
Adi added, “I learned many new things about statistics I didn’t even know before. Learning about these topics really opened my eyes into how big the world of statistics can be. It exposed me to how this field can be used in a number of broad disciplines, from policy to science.”
Twelfth-grader Jennifer Luo analyzed the critical determinants of stroke occurrence under the mentorship of Leah Ung, a junior specialist at the University of California, San Francisco. Her study used stroke data from Kaggle consisting of 5,110 subjects and including variables such as high blood pressure, cholesterol, smoking, obesity, and diabetes. Luo used decision trees and random forest models to find factors that affect stroke occurrence. Her random forest model had better precision performance and identified gender, heart disease, and marital status as significant factors. She also acknowledged that her data was imbalanced, with more non-stroke individuals than stroke, which may explain the lower performance at finding true positive results. Luo suggested improving the accuracy of her model by training it on more cases of stroke.
Discussing her experience, Luo said, “By utilizing resources that my mentor recommended, I was able to learn how to create a decision tree model and a random forest model with my data. The process was challenging and extensive, but I learned a lot.”
Twelfth-graders Adrian Shin and Prajwal Dacharla investigated the impact of environmental pollutants and lung cancer rates across US counties. Advised by Tao He, a professor at San Francisco State University, and former SFASA student Anurag Jakkula, now a freshman at Columbia University, Shin and Dacharla examined data from Harvard Dataverse, a public data repository, which detailed the harmful pollutants in the air in counties across the nation and their respective lung cancer rates. Using their data, they found fine particles, inhalable particles, air quality, and sociodemographic indices were important variables to consider for lung cancer rates. They plan to expand their project by implementing various neural networks to obtain greater precision.
Shin said, “The SFASA project not only sharpened my technical abilities but also reinforced my passion for data science, especially in fields like health care, where data can drive positive change. It was an incredibly rewarding experience that deepened my interest in pursuing future opportunities to use data for meaningful impact.”
Dacharla added, “The collaborative aspect heightened my communication skills. This experience has considerably benefitted my portfolio and robustly fueled my passion for data science. My gratitude towards my fellow peers, the mentors, and the SFASA cannot be fully expressed through words.”
Twelfth-grader Pingyao Liu investigated statistical models for lung cancer prediction. Under the guidance of Senior Principal Statistical Scientist Lijia Wang of Genentech/Roche, Liu focused on exploring lung cancer prediction, aiming to identify trends and factors that could influence the diagnosis. This process involved a mix of data cleaning, exploratory data analysis, statistical modeling, and interpretation of results. Liu used univariate and multiple regressions in R to find significant variables in predicting patient survival. Based on the multiple linear regression, Liu and Wang determined gender and hypertension were the most important factors to consider in the diagnosis process. They also validated their model using the mean squared error score 2.760522, a low value that signifies more accuracy. In the future, Liu hopes to train her model with data from actual patients.
Liu said, “Through this program, I realized that I want to advance scientific knowledge but also make a tangible impact on people’s lives. The opportunity to work with Lijia has provided me with invaluable insights into the daily work of a biostatistician, allowing me to appreciate the power of data science and to see its implications for future medical research.”
Eleventh-grader Anna Khodakovskaia analyzed EEG data recorded during arithmetic tasks. Mentored by Anandamaye Majumdar, a professor at San Francisco State University, the returning student was eager to continue with the program. With the EEG data from 36 test subjects performing arithmetic tasks, she found brain activity varied across different channels or brain regions. Although she found no universal results, she identified brain regions in which brain wave power (such as power of alpha waves and beta waves) showed a significant difference during the mental task for most participants. In the future, she hopes to use the more subject-specific variables of sex, age, and number of subtractions as other clustering methods to find trends in brain activity.
About her experience, Khodakovskaia said, “I found it especially interesting to try to compare the results with what I’ve learned about the brain, seeing where they aligned and where they differed from my expectations.”
Eighth-grader Anondo Deepro Chowdhury presented his findings on statistics from the 1993 World Cup. The youngest student in the program looked at sports statistics from the 1993 World Cup to determine the probability of a team winning a particular match. These statistics included win-loss records and goals scored. With the help of Alexandra Piryatinska, a professor at San Francisco State University, Chowdhury was able to find several relationships in the data and ways to predict wins and losses for teams in the 1993 World Cup. In the future, he hopes to take on more advanced and intriguing statistical projects.
“Overall, the experience was very educational and has gotten my interest in exploring statistics,” said Chowdhury “I love the balance of complexity and learning the project provided, and I am eager to take on more statistical challenges in the future.”
The SFASA plans to host the program again next summer, possibly having in-person components compared to the all-virtual programs of past years. Former students can become mentors in future years, fueling their passion for statistics while inspiring new students.
Leave a Reply