A Problem-Based Introduction to Machine Learning in the Undergraduate Organic Chemistry Laboratory: Prediction of Diels–Alder Reaction Rates
Ricky Tran, Valerie Brunskill, Amanda Musgrove, Todd C. Sutherland, Darren J. Derksen
Abstract
To engage students in higher-order thinking skills, an inquiry-based dry laboratory experience was developed for upper-year undergraduate students, where students were introduced to machine learning approaches to solve chemical problems. Students constructed their own data set of diene and dienophile features and performed a multivariate linear regression in a Python environment to predict the energy barriers (Δ G ‡ ) of a Diels–Alder system. They applied their models to a simulated drug development problem. Likert-scale surveys and qualitative interviews were utilized to collect data on student experiences. Students expressed that they felt strongly engaged in critical and creative thinking, collaboration, and metacognition. Subsequently, students felt that computational tools were more approachable, and had a stronger appreciation of how computational tools could be utilized in chemistry contexts. Students also expressed that they felt the freedom to make mistakes, reflect, and improve, embracing a growth mindset in this laboratory.