Search for: All records

To broaden participation and diversity in data science, educators are increasingly leveraging the adaptation and sharing of successful course models. This paper presents our experience implementing a foundational data science course, adapted from the University of California, Berkeley's Data 8 Foundations of Data Science, at Northeastern University's Oakland Campus in Spring 2024. A key objective was to cultivate student engagement and demonstrate the relevance of data science across disciplines. We assessed the impact of this adaptation on a cohort of first-year students, all non-data-science majors with limited prior programming or statistical experience. Our evaluation focused on student engagement, academic trajectory, and the course's ability to spark sustained interest in data science. The results demonstrate a significant positive impact: 44% of students declared a major in data science or a combined major (e.g., data science and business or economics), 16% pursued a minor in data science, and 16% transitioned to computer science. These outcomes emphasize the importance of designing introductory data science curricula to serve diverse student populations. By incorporating real-world applications from health, economics, social sciences, entertainment, sports, and finance, students gained a deeper understanding of the field's potential and their own capacity to contribute. Furthermore, smaller class sizes promoted interactive learning and personalized assignments, creating a more engaging and accessible educational experience. This approach effectively strengthens students' comprehension of data science pathways and cultivates motivation, ultimately contributing to a more inclusive and diverse data science workforce. 

Abstract Unblinded sample size re‐estimation (SSR) is often planned in a clinical trial when there is large uncertainty about the true treatment effect. For Proof‐of Concept (PoC) in a Phase II dose finding study, contrast test can be adopted to leverage information from all treatment groups. In this article, we propose two‐stage SSR designs using frequentist conditional power (CP) and Bayesian predictive power (PP) for both single and multiple contrast tests. The Bayesian SSR can be implemented under a wide range of prior settings to incorporate different prior knowledge. Taking the adaptivity into account, all type I errors of final analysis in this paper are rigorously protected. Simulation studies are carried out to demonstrate the advantages of unblinded SSR in multi‐arm trials.