Build a Real-Time NYC Taxi Dashboard in Microsoft Fabric

Seventeen labs built a platform. But a platform is not a deliverable: nobody outside the data team ever opens a lakehouse. What the business sees is at the very end, a report and a dashboard. Module 18 of the DP-700 course builds both.

Part 1: A real-time dashboard on the eventhouse

Create a real-time dashboard named Taxi live in Taxi Dev. A real-time dashboard is made of tiles, and every tile is a KQL query with a visual.

The question of the first new tile: which zones are busiest right now? From trips_live, the last few minutes, count trips by zone, top ten.

Save, then Refresh: the bars move, because trips keep arriving. Under settings, a dashboard can refresh itself on a schedule.

Part 2: The measures the report needs

In the semantic model, the report needs measures beyond simple sums, for example the share of trips paying the congestion fee: DIVIDE(CALCULATE(COUNTROWS(fact_trip), fact_trip[congestion_fee] > 0), COUNTROWS(fact_trip)).

Part 3: Three pages, three questions

Demand: where and when? Trips per day, by hour and by borough.

Revenue: what does a trip earn? Per trip, per mile, by time of day, and by payment type.

Congestion charge: what did the 2025 charge change? Fee per day, the share of trips that paid it, and where.

How you know it worked

The workspace lineage shows the whole platform: landing, bronze, silver and gold, the notebooks and pipelines between them, the semantic model, the report and the real-time path.

Two front ends, two questions

• The report, on the semantic model and the warehouse: months of history, as fresh as the last load.
• The real-time dashboard, on the eventhouse: what is happening in the last few minutes.

Knowing which one a question needs is exactly the kind of choice the exam tests.