Six projects built on real data. Each one states a finding, shows the work behind it, and explains why the dashboard looks the way it does. The datasets run from 7,000 rows to 12.6 million.
Excel · Tableau · SQL · Python · pandas
Nearly half of everyone the company acquires is gone within a year.
7,043 customers split into four tenure personas. The headline churn rate of 26.5% hides a first-year rate of 47%, and the mechanism turns out to be contract type.
Read the project →175 games are rated as highly as the famous ones and reach one twenty-first of the players.
82,956 games, labelled and compared in Excel. Same price, same rating, a fraction of the audience. Includes a cut I ran, disproved, and threw away.
Read the project →581 films out of 12.6 million titles clear every bar.
IMDb's full catalogue, filtered down to films rated 8.0 or better with enough votes to trust the score and few enough to be unknown.
Read the project →7,574 songs that listeners replay and radio never played.
Listening data collected from the Last.fm API, joined in SQL to 68 years of Billboard chart history across 1,570 charting artists.
Read the project →A 56-task migration plan where no date is ever typed twice.
Eight stages, eight sign-off gates, and the seven artefacts that make up the paper trail. Generated from code so the plan and the gate sheet can never disagree.
Read the project →Eleven skill kits and forty-plus how-to guides, written for people new to the work.
Built to answer the questions I had when I started. It is here as evidence of the other half of the job, which is explaining the work to people who do not do it.
Read the project →Each write-up follows the same shape, because the shape is the point.
| Section | What it answers |
|---|---|
| The finding | What the data says, in one sentence with a number in it |
| The question | Both possible answers, named before the data decides |
| The obvious objection | The first challenge anyone would make, answered |
| What I threw away | Cuts I ran that did not survive, and why |
| What I checked | The limits of the result, stated before anyone asks |
| Why it looks like this | Every design choice, what I turned down, and why |
The last three do the real work. Charts show that I can use a tool. What I discarded, what I checked, and why I made each choice are the parts that show judgment.