Providing hands-on training in the use of scientific software is a key component of Unidata's service to the geoscience education and research community. Three members of the Unidata Program Center staff recently took part in teaching an American Meteorological Society Short Course on Python for Climate and Meteorology, held virtually over four half-day sessions March 2, 4, 9 & 11, 2021.
Unidata Python developers Ryan May and Drew Camron, aided by educational designer Nicole Corbin, taught the third of the course's four sessions, focusing on Unidata's MetPy package. Before the MetPy session, learners were introduced to fundamental data analysis concepts including package management, functions, GitHub, and manipulation of large datasets, all with a slant toward tools and techniques of use in the atmospheric and ocean sciences. Building on that solid foundation, the MetPy segment helped learners obtain, prepare, and visualize current weather data in a multi-layer map. The final course session let learners apply their new MetPy skills while working with the popular Py-ART and Pangeo packages.
Despite the limitations of the virtual format, the course was fully subscribed, with 48 registered participants, 21 of whom were from US educational institutions. Unidata's team was invited to participate in the workshop by the other organizers, who included Damien Irving of the University of New South Wales, Kathy Pegion of George Mason University, Sarah Murphy of Washington State University, Scott Collis of Argonne National Laboratory, and Kevin Tyle of the University at Albany, State University of New York.
While Unidata's Python development team has taught many MetPy workshops in the past, they have primarily been delivered face to face in a classroom. Adding Corbin's experience in educational design to the mix allowed May and Camron to update both the content and structure of the in-person MetPy course for a virtual environment and a larger class size.
The workshop was conducted using the AMS-provided GoToTraining platform, which provides text chat and “breakout room” functionality. Learners worked through course materials in a Jupyter notebook and submitted answers to the activities through the chat interface. For this session, the team minimized the use of breakout rooms and instead focused on using the Slido online polling app to gauge participants' learning through surveys and multiple-choice questions.
While the Jupyter notebook used for this workshop was created for a guided learning experience, it contains contextual information and activities that should allow users to gain knowledge from the exercise even without direct interaction with an instructor. Following some slight modifications to allow it to work better as a stand-alone resource, the notebook notebook will be available as part of the Unidata Python Training resource. In the meantime, you're welcome to peruse the notebook with solutions that were provided to students during the workshop.
Unidata's software training workshops are usually run entirely by the instructors (sometimes with assistance from professors or students from the hosting institution). Having a third Unidata staffer on hand increased the team's ability to evaluate how well the session achieved its goals. “We evaluated the session on three levels: reaction, learning, and perceived behavior change. In the post-class survey results, we found that the learners enjoyed coming to the course and recognized the coordination and level-of-effort required to put on a workshop like this one,” says Corbin. “The learners also reported feeling comfortably challenged by the material, and we reported a high percentage of correct responses to our assessment questions. When asked, the learners self-reported that they expect to use the skills and knowledge they learned in the session in their future work.”
While workshop participants and Unidata instructors alike thought the course came off well, the virtual format did present its own hurdles, including getting all of the learners' systems configured and working properly. “Debugging remote environment issues, across all the different OS permutations and different levels of user permissions” was an expected challenge, says May. “Doing it quickly so that learners don't fall behind or so there's not a significant impact on our schedule is tough and can be stressful.” Still, Unidata staff are hopeful that building our capability to present material in a virtual setting will let us reach a wider range of potential learners and foster connections similar to those formed at in-person events.