I never intended to become a data science researcher. It started as a joke between my flatmate and me during exam season. We were both stressed, living on instant noodles and nervous energy, and one evening I said, “Wouldn’t it be great if we could just predict our grades before we even sit the exam?” We laughed, but the idea stuck with me. The next morning, I opened a blank spreadsheet and started collecting data — how many hours we studied, how many past papers we attempted, how much sleep we got the night before. It was messy and completely unscientific, but watching those numbers turn into patterns — even unreliable ones — felt strangely addictive. I had stumbled into data science through the back door, wearing pyjamas and fuelled by anxiety.
When our predictions turned out to be wildly inaccurate (I scored much higher than expected on one module and disastrously low on another), I wasn’t disappointed. I was intrigued. What variables had I missed? What if I had included lecture attendance, or the time of day I studied, or even my mood? The realisation that I could frame a question, gather data, and test an idea — even badly — was electrifying. Data science, I understood for the first time, wasn’t just about building flawless models. It was about asking questions and learning from the answers, including the messy, uncomfortable ones.
By the time dissertation season arrived, I knew I wanted to explore something that combined this hands‑on curiosity with a more rigorous academic approach. But I needed a topic that was ambitious enough to stretch me without being so complex that I’d drown. I spent a rainy Saturday browsing through data science dissertation topics and it was exactly the guidance I needed. Some students had built predictive models for student dropout rates, others had applied machine learning to climate data, and a few had used natural language processing to analyse mental health discussions on social media. Seeing that variety gave me the confidence that my own interest — in using machine learning to understand factors affecting student academic performance — could become a legitimate, researchable project.
With my focus sharper, I began to design my study. I decided to collect anonymised data from students across multiple departments, including demographic information, study habits, and prior grades, and to build a classification model to predict end‑of‑year outcomes. My supervisor helped me navigate the ethical approval process and pointed me toward relevant literature on educational data mining. The project felt personal, grounded, and genuinely useful — not just an academic exercise but something that could, in a small way, help future students understand their own learning.
If you’re drawn to data science but feel intimidated by the technical complexity, start with something you already care about. Maybe it’s a question about your own university experience, a community problem you’ve noticed, or a dataset from a hobby you love. The best topics often grow from everyday curiosity — not from a desire to sound clever, but from a genuine itch to understand something better. Then explore what other students have already done, and use that to sharpen your own idea. You don’t need to be a machine learning expert from day one. You just need to be curious enough to ask a good question, and brave enough to try answering it. The rest will follow.