Applied Statistics and Data Science, MS postgraduate personal statement example

Personal statement example

Every summer Saturday I sit in a wooden hut beside a village cricket pitch with a pencil and a scorebook. I have scored for our club's second eleven for six seasons, and at the end of each year I type the books into a spreadsheet so the captain can see batting averages. Two years ago I noticed that our most praised opener averaged well below a player who rarely got picked, and that the difference largely disappeared once I separated innings played on our sloping home ground from those played away. Nobody had been wrong, exactly; they had been looking at a single number that hid where the runs were made. Questions like that, about what a summary conceals and how to account for it honestly, are why I want to study applied statistics and data science at postgraduate level.

My degree was in Geography, and the most useful part of it turned out to be quantitative. For my dissertation I used publicly released daily trip counts from a city cycle-hire scheme alongside weather records to ask how rainfall and temperature affected usage. My first attempt was ordinary least squares regression, but the residuals made it plain that counts behaved differently from the model's assumptions. I taught myself enough to fit Poisson and then negative binomial models in R, using day of week and school holidays as additional predictors. The final model was modest, and I was careful in my write-up to say that it described association in one city over three years rather than anything more general. My supervisor's main comment was that my limitations section was the strongest part, which I took as a fair compliment and also as a sign of how much I still had to learn about model checking.

Since graduating I have worked as a stock controller at a distribution warehouse for a homeware retailer. Much of the job is cycle counting: checking a sample of locations each day against the system's recorded quantities. When I started, locations were chosen more or less at random by whoever was on shift. I suggested weighting the daily selection towards product lines with high pick rates and a history of discrepancies, and built a simple spreadsheet that ranked locations accordingly. My manager agreed to trial it, and over the following months our counts found discrepancies earlier, before they caused failed picks. It is not sophisticated work, but it showed me that a sensible sampling decision can matter as much as the analysis that follows, and that people will adopt a method only if they understand why it works.

Outside work I have been filling gaps in my preparation. I have worked through linear algebra and probability material using free university lecture notes, and I read Charles Wheelan's Naked Statistics and then the more demanding An Introduction to Statistical Learning, working through its R labs. The chapter on resampling methods was the first time cross-validation made intuitive sense to me rather than being a procedure to follow. I am now more comfortable with Python too, mainly through rewriting my dissertation analysis in pandas and statsmodels to compare the two workflows.

I am applying because I have reached the edge of what I can do responsibly by teaching myself. I can fit a model, but I want a firmer grounding in inference, experimental design and machine learning, and the discipline that comes from having my reasoning examined by others. In the longer term I would like to work in logistics or public transport planning, where decisions rest on messy operational data and where clear explanation to non-specialists matters. I bring practical experience of working with imperfect records, a habit of stating limitations plainly, and the patience that comes from scoring a forty-over match with a broken pencil sharpener. I would welcome the chance to build on those foundations properly.