Senior Data Scientist
2018 - 2021SimIS- Improved learner behavior prediction accuracy from a 21% baseline (recommended next action) to 66% on unseen test data via a long short-term memory recurrent neural network (LSTM RNN) model.
- Predicted course completion with Matthews correlation coefficient 0.51 using Experience API (xAPI) student log data and built the corresponding explanatory model via factor analysis.
- Co-authored an e-learning metadata analytics strategy distributed across the Department of Defense and chaired the stakeholder working group on its adoption and implementation.
Technologies: SQL, Machine Learning, Artificial Intelligence (AI), PythonBusiness Analyst
2012 - 2013Accenture- Eliminated test case redundancies via Excel data analysis, cutting testing time by nearly 15%.
- Led deliverable review sessions with high-level stakeholders across multiple teams to ensure business requirement compliance.
- Performed ad hoc defect analysis to facilitate efficient prioritization of cross-functional effort.
Technologies: Tableau, SQL
Yihua Liu
Data Science Developer in Orlando, FL, United States
Member since July 28, 2021
Yihua is a lead data scientist with over a decade of experience across various companies and teams. With several industry journal publications, speaking engagements, and extensive client-facing experience, he enjoys sharing and discussing his work with audiences of all backgrounds, including C-suite executives and non-technical stakeholders.
Yihua is now available for hire
Experience
- Statistics 10 years
- Tableau 8 years
- Analytics 8 years
- SQL 7 years
- Machine Learning 6 years
- Big Data 6 years
- Artificial Intelligence (AI) 5 years
- Python 5 years
Availability
Part-time
Preferred Environment
Python 3, SQL, Machine Learning, Analytics, Artificial Intelligence (AI), Tableau
The most amazing...
...and highest-impact project I've worked on is Covered California, the state of California's health insurance marketplace.
Employment
Experience
- Educational Outcomes Prediction
This project examines educational data—including student demographic information and academic records, school attributes, and teacher data—from kindergarten through third grade for a diverse cohort of students.
In the exploratory phase, we find methods to reduce the minority achievement gap and improve all students' outcomes. Next, we attempt to predict future test scores via several regression models. Finally, we predict whether students will graduate from high school and whether they will take a college entrance examination—SAT or ACT.
Although these events occur nearly a decade after third grade for most students, we were able to perform relatively well, with ROC AUC (area under the curve) scores between 0.7 and 0.8 on unseen test data.
Skills
Languages
SQL, Python 3, PythonOther
Machine Learning, Analytics, Artificial Intelligence (AI), Mathematics, Applied Mathematics, Statistics, Big DataTools
Tableau
Education
- Master's Degree in Statistics2015 - 2016University of Central Florida - Orlando, FL
- Bachelor's Degree in Mathematics2008 - 2011University of California, Berkeley - Berkeley, CA
