Announcing a new eLearning series available now on Unidata eLearning: Supervised Machine Learning Readiness. This learning series is a self-paced, beginner-friendly program designed for Earth systems scientists to explore the core principles of supervised machine learning. This series uses a combination of step-by-step frameworks, exploratory widgets, and low-code exercises in Jupyter Notebooks, to explore the full cycle of machine learning model development. No programming experience is required. By the end of the series, you will be able to recognize when machine learning is an appropriate tool and critically evaluate machine learning in Earth systems science contexts.
Enroll in Supervised Machine Learning Readiness with your free NSF Unidata eLearning account.
This work is supported by the US National Science Foundation under CyberTraining award #2319979 and is developed in collaboration with MSU Denver.
More about the series
This series of modules is designed to enable learners with limited formal education in advanced math, statistics, and programming to be ethical and savvy users of machine learning tools and outputs. By focusing on accessible learning for this audience, we bolster the professional skills of more than just future machine learning developers. We focus specifically on the growing demand for the skills to critically judge machine learning outputs. The goals of this program are to elucidate the conceptual mechanisms behind machine learning models for an Earth Systems Science audience, and bridge the gap between machine learning conceptual mechanisms and "low-code" real-world applications for inquiry into Earth Systems Science processes. We assess this program on the following four aspects:
- The learner's ability to assess the suitability of using machine learning for new Earth Systems Science problems
- The learner's ability to accurately describe a previously completed Earth Systems Science machine learning study in terms of the data, model, and evaluation techniques used as well as the impacts the outputs have on the research scenario
- Given a new Earth Systems Science problem in which machine learning is needed, the learner's ability to create, execute, and justify an appropriate project plan that includes general steps for data preparation, model choice, evaluation techniques, and expectations for outcomes
- The learner's ability to describe the impact of the acquired technical and critical thinking skills on their success with advancing in their desired career
The full series of three modules:
- Machine Learning Foundations for Earth Systems Scientists: No-code conceptual introduction to supervised machine learning in the Earth Systems Sciences. Learners explore problem framing, data handling, and model development concepts with support from no-code widgets.
- Machine Learning Applications for Earth Systems Scientists: Low-code exploration of concepts in Module 1. Learners create and test hypotheses using Jupyter widgets and pre-populated code. Learners work with real-world data and test the effect of various preprocessing and training strategies.
- Machine Learning Analysis for Earth Systems Scientists: Team-based lab module. Groups of students select a scenario to create and execute a machine learning strategy for. Groups must justify the choice of data, model, and other decisions made for their scenario.
