Think ML Is Too Hard? STEM Students Are Already Halfway There

I first considered Machine Learning as something challenging yet feasible for me to learn when I attended the American Astronomical Society conference in Florida in 2016 and saw a fellow UIUC undergrad senior present research on Machine Learning in Cosmological Simulations. Partially because he had a reputation for prioritizing research over sleep and also because he got into grad school at Harvard, I dreaded that only the brightest and most dedicated individuals could handle Machine Learning.

In my first year of grad school, I found and downloaded a PDF related to Machine Learning with Python, but I kept putting off diving into it until my last semester of grad school when I had decided to transition to data science after my PhD. When I finally learned ML that semester, I was surprised there wasn’t more to it. Of course, there is always more to learn, but I wished I had just spent a little time years earlier to learn the basics.

Acquiring a basic knowledge of Machine Learning as well as of the pandas library in Python would have made me a much better and more efficient researcher during my grad student years. Even just knowing how to explore data systematically, fit regression models quickly, and test hypotheses computationally would have greatly accelerated my pace of work.

1. The Math of ML Isn’t as Scary as It Sounds

The math behind machine learning is primarily from linear algebra, calculus, probability and statistics, and optimization methods. As an undergraduate student in a STEM field (Science, Technology, Engineering, or Mathematics), you probably already take several of these courses as prerequisites for your degree.

I never took a formal statistics course, but I did learn some statistics during my courses in thermal physics and statistical mechanics, as well as during my research. You’ve probably already learned least squares optimization, know some linear algebra, and understand how to construct linear models. That background is enough to recognize the logic behind most machine learning algorithms.

It’s worth noting that you don’t need to master every piece of the math before starting. Most of the heavy lifting is already built into libraries like scikit-learn. What matters most is developing intuition: when to use which model, what assumptions are baked in, and how to interpret the results.

2. From Coding Labs to ML Models: The Skills Carry Over

Many STEM students already learn computer programming in some courses (from my experience courses like intro to computing, physics labs, data structures, and numerical methods) and during research. If you can write code to solve a differential equation or process data from a lab instrument, you have the skills to write Python scripts that run ML algorithms.

The barrier to entry is lower than ever. Jupyter notebooks make it easy to prototype and visualize results. Pandas streamlines data cleaning and manipulation. Kaggle and open datasets give you free, real-world problems to practice on. Even if you haven’t coded extensively, today’s ML ecosystem is designed to be approachable.

3. Bridging the Gap

Machine Learning can sound intimidating because of the buzzwords, but many core ideas are straightforward extensions of concepts you already know. Of course, there are more complex applications—natural language processing (NLP), computer vision (CV), and large language models (LLMs). But you don’t need to start there, or even get to them at all, to benefit from ML. A foundation in regression, classification, and clustering already unlocks a huge set of practical tools.

The payoff for learning even the basics is enormous. In research, you can automate data analysis, fit models faster, and uncover patterns you might miss by hand. In industry, ML skills are among the most in-demand and transferable across fields. In education, ML literacy helps you critically read modern scientific papers, many of which already rely on these techniques.

You don’t need to dedicate years to get value from ML. A few weekends of focused effort can give you enough to start applying it to your own problems.

If you’re a STEM student, don’t wait like I did. You already have the mathematical and computational foundation to succeed. Spend a little time learning ML basics—you’ll be surprised at how accessible it is and how quickly it pays off, whether you continue in academia or move into industry.