Then I started looking at how real businesses actually run.
The gap
Most of them sit on data they never use.
So I moved from economics into business analysis and data science, to close that gap.
First real fix
Month-end close, 30% faster.
As an accounting intern at Basil & Alred in Tanzania, I automated the monthly reports in Excel and rebuilt the dashboards leadership read.
What will this ride cost?
$2
average miss when it predicts a fare
22,699 trips full of gaps and odd outliers. I cleaned them and built a model that predicts the fare before the ride. Each dot is one trip: the closer to the line, the better the guess.
Practice projectPython · scikit-learn
real fare →predicted fare →
Four more messes, four answers.
Same habit every time: messy data in, a decision someone can make out.
Is this company healthy?
90%
less time to build a report
Judging a company means working out dozens of ratios from its financial statements. My app does it straight from the Excel files, compares companies side by side and flags warning signs in red. Here, Boryszew carries the most debt of the three.
Patients, treatments and payments lived in separate files, so every question meant matching them by hand. I put them in one database with a live dashboard, so managers can see revenue month by month on their own.
Airlines pay penalties for late flights, but a huge flight log doesn't tell you where to look. My Excel dashboard filters delays and penalties by airline and state in one click, and points straight at California.
Practice projectExcel · Power Query
Which EU countries are alike?
4
groups of look-alike countries
Comparing 27 countries on six measures at once is too much for any table. I pulled World Bank data and let an algorithm sort the countries into groups that behave alike, in a chart you can explore.