(506c) Tips and Pitfalls to Avoid When Teaching Machine Learning with Python to Chemical Engineering Students at the Undergraduate and Graduate Level

Every area of Chemical Engineering research and industry, including Catalysis, Biomolecular Engineering, Cellular Engineering, Computing and Simulation, Nanotechnology, Materials, Polymers, Sustainable Energy, and Microfabricated Systems, stands to be transformed by data science—the ability to produce, mine, build models, and learn from vast quantities of data. However, many chemical engineers have little to no training in data science and machine learning. In the Department of Chemical Engineering at University of Michigan, I have recently created and taught two classes on Data Science and Machine Learning to graduate and undergraduate chemical engineering students. The classroom material is aimed to prepare students to use modern data science tools (e.g., Python programming and the Numpy, scikit-learn, matplotlib, and pandas libraries) to analyze data that engineers may encounter. Course content expands the students’ horizons of the utility of data science in real-world context and enhances written and oral communication abilities as well as teamwork. Topics and applications covered include data collection, curation, and supervised and unsupervised machine learning. Algorithms covered will include the perceptron, adaline, logistic regression, gradient descent, principal component analysis, random forests and feed forward neural networks. Homework and lab exercises include hands-on practice of using data science and machine learning with Python in Google Colab notebooks. Students are responsible for a team-based data science / machine learning project. In this talk, I will share my experience and provide some tips and pitfalls to avoid when teaching data science / machine learning content to chemical engineering students.