Sowiso Performance Analysis — Excel Project

View PDF

Students compared

Math sub-themes

100

My score

88.6

Class average

1The setup

In my first week of the BSc Business Analytics program at Maastricht, all 81 of us completed math homework on the Sowiso platform. The system records two things for every exercise: score (out of 100) and duration (seconds spent).

I wanted to see where I stood. Not just my own grade — but how I compared to the rest of the class across 64 mathematical sub-themes like "the notion of functions", "the equation of a line", "completing the square".

      The question: does first-week Sowiso performance actually tell us who's good at math? And where do I rank?
    

2The approach

I exported the data into Google Sheets and used the =AVERAGE() function to compute, for each of the 64 sub-themes, the class mean for both score and duration. Then I built three bar charts:

The class average score per sub-theme (one bar for each of 64 topics)
My own score per sub-theme (also one bar each)
A side-by-side comparison of my totals vs the class totals

3Result #1 — the score

Across all 64 topics, I scored a perfect 100. The class averaged ~88.6.

SCORE COMPARISON

So mathematically — yes, I crushed the homework. But this was already starting to feel like an unimpressive metric: a class average of 89% means almost everyone is acing it. Sowiso's first week is the easy onboarding round, not a true skill test.

4Result #2 — the embarrassing part

Looking at the duration, things got interesting:

DURATION COMPARISON

Class average: 447 seconds per topic. Mine: 2,568 seconds — almost 6× longer.

      The outlier: on one specific topic ("Functions and graphs 4"), my duration was 89,321 seconds. That's 24 hours straight. Obviously, I'd left the browser tab open overnight 🙃 — but it was enough to single-handedly skew the class average for that topic.
    

This was the moment I realized something important about data analysis: raw aggregates lie. One careless data point can make an entire group look slow. The class average of 447 seconds itself was inflated by people like me leaving tabs open.

5Conclusions

This project was tiny in scope but taught me lessons I still use:

Always check for outliers before trusting an average. A median or trimmed mean would have told a much truer story.
"Performance" isn't one-dimensional. Score and duration interact — a high score from someone who took 6× longer means something different than the same score from someone who took 5 minutes.
The first-week Sowiso doesn't measure math skill, it measures attendance. The real signal would come from harder exercises later in the year.
Data tells stories — but only if you ask the right questions. "Am I above average?" is the wrong question. "What does 'average' even mean here?" is the right one.