Teaching Machine Learning to Materials Engineering Students Using Free, No-Code Software

What is it about?

This study evaluates the use of Orange Data Mining, a free and intuitive platform, to teach the basics of artificial neural networks to materials engineering students without requiring any programming skills. Featuring a drag-and-drop visual interface, the software allows users to construct, test, and manipulate custom predictive models. As a practical case study, the researchers utilized the tool to predict the water solubility of 412 inorganic compounds and the dissolution rates of 19 inorganic glass compositions. Students and educators can easily adjust key model parameters—such as the number of hidden layers, neurons per layer, and activation functions like ReLU—to instantly observe how hyperparameter changes impact predictive accuracy. Ultimately, these visual models demonstrated high accuracy in capturing complex physicochemical relationships from experimental data.

Featured Image

Photo by Steve A Johnson on Unsplash

Photo by Steve A Johnson on Unsplash

Why is it important?

Machine learning is transforming materials engineering by enabling researchers to predict material properties before physical synthesis, saving considerable time and laboratory resources. However, traditional machine learning frameworks like TensorFlow and PyTorch rely heavily on complex coding skills and deep mathematical backgrounds, creating a significant barrier for undergraduate students entering the field. Platforms like Orange democratize access to these predictive techniques by providing an interactive, visual approach to complex concepts such as overfitting and hyperparameter tuning. By serving as an accessible educational bridge, this approach helps students build a solid conceptual foundation before moving on to advanced, code-based data science environments.

Perspectives