.pbix file, including the capstone. This page exists because Power BI is the tool you are
most likely to meet at work, and fifteen minutes of hands-on time now will make that first encounter
familiar instead of new.
Power BI is Microsoft's business-intelligence tool. You point it at data, drag fields onto a canvas, and get charts, tables, and filters a non-technical person can operate. At most firms it is the layer on top of the analytics work you have been doing all term. The model runs on a schedule, writes its scores to a table, and Power BI serves those precomputed numbers to managers as a dashboard. BI over batch-scored data. The Week 11 enrichment deck's dashboard-tools tour walks through this same picture, with an Excel prototype alongside the Power BI segment, so treat this page as the hands-on companion to that deck.
The skill underneath is the one you are practicing this week anyway. Wire a model's output to an interface someone else can operate. The tool changes from job to job. The skill does not.
| Power BI Desktop | Free. Windows only. Install from the Microsoft Store or Microsoft's download page. This is the authoring tool, and it is all you need for this page. |
| On a Mac | There is no native Mac version of Power BI Desktop. If you are on a Mac, read this page for the concepts and skip the try-it, or run it later on a Windows machine. You lose nothing gradeable. |
| The paid part | The paid line sits on the publish-and-share side. Building on your own machine is free; distributing dashboards to other people inside an organization is where licensing starts. Details shift year to year, so keep the coarse version: authoring free, sharing paid. |
You already know this dataset. It is the Project 2 credit-default data,
ar_default_data.csv, and bad_ar is the default flag.
https://raw.githubusercontent.com/sean-mccaman/acctg5150-090/main/2026-summer/project-2/ar_default_data.csveducation to the X-axis and
bad_ar to the Y-axis. Power BI will sum it by default; open the field's dropdown in the
Y-axis well and switch the aggregation to Average. The average of a 0/1 flag is the default
rate, the same trick you used in pivot tables. You now have default rate by education segment.marital_status into it. Click a value.
The bar chart refilters instantly. Verify what you built: pick one slicer value and tie one bar out
against a pivot-table average of bad_ar on the same filter. Congratulations, you have made
the thing that is on half the office monitors in corporate America.That slicer filters precomputed data. Every number it can ever show was already in the table when you clicked Load. That is completely normal at work. It is also below the capstone bar. The capstone asks for recompute at interaction time: a control that re-scores through the model's coefficients, or a threshold control that recomputes flags, precision, recall, and expected cost from the score vector. A slicer over a saved results table is the one way to fail that bar. Know which of the two you are building, and you will never be surprised by the distinction again.
Power BI can do true recompute. A What-if parameter gives you a slider whose value feeds a DAX measure, and a measure recalculates on every interaction. Write the threshold as a What-if parameter, write flagged count, precision, and recall as DAX measures over the score column, and the dashboard recomputes at interaction time. That is the real thing, and building it once is a genuinely useful rep. Do it for yourself if you want it.
Either way, .pbix is not a submission path in this course. The grading machine cannot open
it, which makes it an ungradeable deliverable. The gradeable equivalent of everything on this page is the
Excel live-recompute path: input cells drive a live coefficient formula instead of a slicer over a
static table. The Week 11 exemplar workbook shows exactly that build, and it earns equal credit with the
taught HTML path.
Back to the Week 10 page.