Materials Informatics (the field that bridges materials science and data science) covers a lot of ground, and the big picture can be hard to see at first. This page helps you first pick an entry point that fits your background, then follow a four-stage path: Overview to Core to Expand to Apply. Every step has a checkbox, and your progress is saved in this browser. The order is a suggestion, so feel free to start wherever your interest lies.
⏱️ Want to try it in about an hour?
If you would rather get hands-on before the theory, start with the shortest path. And if a term trips you up, the glossary is there to help.
🚪 Entry points by background
MI sits where several fields meet. Start by shoring up whichever side feels least familiar for your background, and the rest of the journey becomes much smoother. Once you finish your entry point, join everyone at Stage 1, "Get the big picture," below.
Materials / chemistry background (unsure about ML)
For those grounded in materials or chemistry but new to machine learning. Build the shared ML vocabulary first, then join the main path.
- Take Introduction to Machine Learning to survey the core terms of supervised / unsupervised learning and evaluation.
- Then continue to Stage 1 to get the big picture of MI.
Informatics / data background (unsure about materials)
For those strong in machine learning and programming but uneasy with materials-science vocabulary. Cover the materials entry point first.
- Take Introduction to Nanomaterials to pick up the concepts and vocabulary of materials science through hands-on Python.
- Then continue to Stage 1 to get the big picture of MI.
Students / building from the math up
For those who prefer to solidify the mathematical foundations in order. Probability, statistics, and linear algebra underpin nearly every MI method.
- Take Probability and Stochastic Processes to gain the language for reasoning about uncertainty.
- Take Linear Algebra and Tensor Analysis to solidify the math behind data and features.
- When you are ready, move on to Stage 1.
📈 The staged learning path
From the big picture to real applications, in four stages. Use each step's checkbox to record your progress (it is stored only in this browser and is not shared with anyone). Chapter counts and time estimates are approximate, based on the actual materials.
Get the big picture
First, take an aerial view of what MI is and how it works. This is where every entry point converges.
-
Introduction to MI You will grasp the mindset of data-driven materials development and how the whole process flows. Open series →
Core techniques
Gather data, turn it into features, analyze it, and optimize with few experiments — the four techniques at the heart of MI.
-
Introduction to Materials Databases You will be able to retrieve and clean materials data from sources such as the Materials Project. Open series →
-
Introduction to Composition-Based Features You will learn to design machine-learning features from composition (Magpie / matminer). Open series →
-
Introduction to Experimental Data Analysis You will be able to analyze experimental data such as XRD and microscopy images. Open series →
-
Introduction to Bayesian Optimization and Active Learning You will learn search strategies that reach optimal conditions in few experiments. Open series →
Expand your reach
Building on the core, broaden into representation learning, simulation, efficient search, and large-scale computation.
-
Introduction to Graph Neural Networks (GNN) You will learn to treat molecules and crystals as graphs and predict properties from structure. Open series →
-
Introduction to Machine Learning Potentials (MLP) You will grasp the basics of next-generation simulation that combines quantum accuracy with classical speed. Open series →
-
Introduction to Active Learning You will learn strategies for choosing "what to measure next" to boost experimental and computational efficiency. Open series →
-
Introduction to High-Throughput Computing You will learn to automate DFT calculations and evaluate large numbers of candidate materials efficiently. Open series →
Take on applications
Connect what you have learned to real challenges: batteries, catalysts, drug discovery, and lab automation. Feel free to pick whichever is closest to your interest.
-
Battery Materials MI Application You will practice data-driven materials design for next-generation batteries. Open series →
-
Catalyst Materials MI Application You will learn how to theoretically design and screen highly active catalysts. Open series →
-
Drug Discovery MI Application You will follow the flow of AI-driven drug discovery, from molecular design to ADMET prediction. Open series →
-
Introduction to Robotic Lab Automation You will grasp the big picture of closed-loop materials development with autonomous experiment robots. Open series →
🎓 Capstone Project
Once you have worked through the four stages, bring what you learned together into a single deliverable. It is a self-directed mini-project where you build a property-prediction model from scratch on data you choose.
Build a property-prediction model on your own data
An assignment template where, on 30+ samples of materials data, you run featurization to a baseline to cross-validation to interpretation to a limits discussion on your own, then write it up in a short report. It comes with a checklist, milestones, a self-assessment rubric, and a report template — turning your roadmap learning into a portfolio piece (about 4–8 hours).
❓ Frequently asked questions
How long does the whole thing take?
Adding up the estimated times for the 13 series in this roadmap gives roughly 25–30 hours in total. At a pace of 30–60 minutes a day, a few weeks to about two months is a reasonable guide. You do not need to finish everything: getting through the entry point plus Stages 1 and 2 already gives you a solid foundation for practical MI work.
How much math do I need?
For the entry point and Stages 1–2, high-school through first-year-undergraduate math (the basics of linear algebra and probability/statistics) is enough to follow along. The material is built to be understandable through code and figures rather than heavy equations. If you want firmer theory, read the Probability and Linear Algebra series from the "building from the math up" entry point alongside. Parts of Stage 3, such as GNNs and MLPs, go deeper when you understand linear algebra.
Can I do this without programming experience?
Each series is centered on Python code examples, and the code is shown in a runnable form throughout. If you are entirely new to programming, start with the Quickstart, typing along to get a feel for the flow, and get comfortable with the basics via Introduction to Machine Learning as needed. If you know basic Python syntax (variables, lists, functions), you can work through the material comfortably.
Do I have to follow the order exactly?
No. The stages are a "recommended order for when you are unsure," and you can start with whichever series interests you. That said, from Stage 2 onward, reading Stage 1, "Get the big picture," first lines up the prerequisites and makes things easier to follow. For the Stage 4 application series, picking just one that is closest to your interest already makes for practical learning.
Where is my progress saved?
Your checkbox state is saved only within your browser (localStorage). It is not sent to any server and is not shared with anyone else. Open the page in the same browser on the same device to pick up where you left off. Note that it does not carry over to other devices. To clear everything, press "Reset progress."