Statistical Foundations in Modern Data Science Education: Insights from a School of Data Science

Music City Center 

Room: CC-104C 

Description:
This session explores the innovative approaches for teaching essential statistical content across undergraduate and graduate data science programs at the University of Virginia's (UVA) School of Data Science. We examine how UVA embeds rigorous statistical thinking and methodologies across courses, embracing its students' diverse backgrounds.

Focus:
The session will highlight UVA's strategies for:
- Integrating key statistical concepts into courses not primarily focused on statistics to students with widely varying mathematical, statistical, and computational backgrounds.
- Maintaining consistent statistical rigor across the curriculum without relying on dedicated statistics courses.
- Emphasizing statistical inference and uncertainty quantification in data-driven decision making.
- Developing a conceptual framework that bridges traditional and contemporary methods, highlighting the evolution from traditional tools to modern approaches.
- Differentiating data science programs from traditional statistics, biostatistics, and computer science programs.

Content:
The session will feature talks from faculty who have developed and teach courses in UVA's data science programs, along with insights from a non-UVA discussant.

Presenters will discuss the following in a coordinated manner:
- The core statistical concepts essential for a data science education.
- Strategies for constructing a curriculum that emphasizes foundational principles rather than a sequence of disconnected tools.
- Approaches to building consensus for a curriculum that spans all domains of data science, including statistics, computer science, engineering, and policy.
- How experimentation and inference have been established as foundational topics in the undergraduate curriculum.
- Innovative approaches to delivering probability and inference concepts to students with widely varying backgrounds, focusing on conveying the core concepts needed to understand uncertainty for reasoning and prediction.
- New methods for organizing statistical modeling and machine learning content, such as reworking traditional linear model material into a broader predictive modeling framework.
- Integrating Bayesian methods and other core statistical concepts across multiple courses.
- The blending of inferential and predictive concepts in data science education.
- How statistical thinking has influenced our content on ethics, fairness, and algorithmic bias.
- Strategies for holistically combining mathematical and computational approaches that maximize student retention and understanding.

Timeliness:
As data science programs proliferate, it is crucial to ensure that statistical foundations remain central to the discipline. This session addresses the timely challenge of incorporating robust statistical content into the rapidly changing discipline of data science, while attracting students from varying backgrounds.

Appeal:
This session will be valuable for:
- Statistics and data science educators seeking to update their curricula.
- Academic administrators involved in planning new data science programs.
- Industry professionals interested in the evolving skill sets of data science graduates.

Data Science

University of Virginia

Education 

Yes

Section on Statistics and Data Science Education

Business Analytics/Statistics Education Interest Group