Unsupervised Anomaly Detection in High-Dimensional Vehicle Telemetry: A Predictive Maintenance Framework for Fleet Operations Utilizing Isolation Forests and Time-Series Analysis

Growing up, my mornings didn't start with an alarm clock; they started with my father and his business routine . My father owns a trucking and logistics business, and for as long as I can remember, our dinner table conversations revolved around one thing: the fleet. I learned very early on that in logistics, a stationary truck is not just a vehicle—it is a liability. I watched my father navigate the immense stress of "unforeseen circumstances." A breakdown wasn't just a mechanical failure; it was a missed delivery window, a penalty fee, a stranded driver, and a hit to a reputation built over decades. The industry was running on experience and gut instinct, but often, that wasn’t enough to predict a radiator failure 500 kilometers away from the nearest service station. When I entered the world of Computer Science, I was introduced to "Predictive Maintenance" as a high-level concept reserved for aerospace or massive industrial plants, typically requiring cloud servers and million-dollar infrastructure. It felt disconnected from the reality I knew. My father’s office didn’t have a data center; it had a dusty desk, a stack of invoices, and a standard mid-range laptop. That disparity became the catalyst for this research. I realized that for technology to be truly useful to the backbone of our economy—the fleet operators—it had to be accessible. It had to be lightweight. It had to work here, not just in a theoretical cloud. This paper, "Unsupervised Anomaly Detection in High-Dimensional Vehicle Telemetry," is not just an academic exercise in dimensionality reduction. It is a personal mission to democratize safety and efficiency. By constraining this project to run entirely on a standard laptop, I wanted to prove that we don't need supercomputers to predict the future of a vehicle's health. We just need the right math, applied with the right perspective. This work is dedicated to my father, who taught me the value of the journey, and to every fleet operator who keeps the world moving, one mile at a time.