Shara Taylor has always liked working with data.
As a research analyst for the Tennessee Housing Development Agency, she ran data for all of the agency’s programs to create a report for the Tennessee’s Congressional members, General Assembly and other interested parties. More recently, she was an urban development specialist at the Metropolitan Development and Housing Agency where her favorite assignments involve pulling demographic and property data for redevelopment districts.
But Taylor knew she was only scratching the surface of what she could accomplished with data.
She began taking completely asynchronous data science classes, but that format did not work for her learning style. She then learned about the new M.S. Data Science program at Meharry.
“I was looking for an online program that covered a wide breadth of classes,” says Taylor. “I also knew that, as an HBCU, Meharry would focus on issues that disproportionately affect the African American population as we explore data.”
She found an online setting that fit into her work schedule featuring live classes with class discussions.
“If I don’t understand something I can ask questions,” says Taylor. “My professors have always been responsive, but sometimes I also learn when other students ask a question I hadn’t thought of yet.”
While the online, evening classes at Meharry SACS make it possible to work while she pursues her degree, she still needs strong time management skills to balance both demands.
“We have class twice a week. I then use my other weeknights and the weekend to complete any assignments,” says Taylor. “I don’t let my schoolwork bleed into my work day because I can’t depend on having that time.”
Even with the program’s flexibility and her strong organizational skills, mastering the curriculum in the data science program requires commitment.
“I do recommend that anyone considering the program really consider their schedule. You really need to be able to commit a significant time outside of class to be successful,” says Taylor.
Over the course of the program, Taylor continues to be amazed by the power of data science tools.
“Python is my favorite software to use. You can do so much and slice data in many different ways to give you a really deep look,” says Taylor.
In class, Taylor has scraped nutritional data from the FDA website and used the programming language R and Excel to build a tool that allows someone to easily develop a calorie count for any meal.
“To make it easy for someone who isn’t a data person to pull that information with just a few clicks is very intriguing to me,” says Taylor. She also applied machine learning to use X-ray images to distinguish COVID-19 from other illnesses affecting the lungs in Dr. Bishnu Sarker’s Computational Machine Learning course.
Using 2,000 images from the COVID-19 Radiography Database, Taylor built a Convolutional Neural Network (CNN) that can distinguish between COVID-19, viral pneumonia, lung infections and normal lungs.
“It was really an interesting project,” says Taylor. “It is amazing that you can build an algorithm that can work at such a granular level and diagnose a disease.”
The M.S. Data Science program concludes with a capstone course, and Taylor’s Natural Language Processing class inspired a project idea that would merge her interest in data with her passion for music.
“We did a project that pulls lyrics from different songs. We then classify the lyrics by content, genre and style so that a Spotify or Pandora can make recommendations to people based on the actual content of the music they listen to versus the genre or the artist,” says Taylor.
Her capstone project could take that project steps further, and also address a problem she finds with the algorithms of music streaming services.
“I love music, but one of my frustrations with steaming services is how they recommend artists,” says Taylor. “I might like a hip-hop artist, but then a service will recommend another artist that sounds similar, but with content I don’t enjoy.”
“I am thinking about a project where I would develop an algorithm that recommends songs by their content. So, after listening to that first hip-hop artist, I might want to hear a pop artist with content covering similar topics,” says Taylor.