📊 Matplotlib, Plotly, or Power BI?

About this simulation

A model in production constantly emits monitoring metrics — accuracy, latency, data drift. The same stream of numbers can end up in three very different places: a Matplotlib figure baked into a periodic PDF report, a Plotly dashboard an ML engineer refreshes live while debugging, or a Power BI report an executive glances at once a week. This scene renders all three stations at once, fed by the same underlying metric stream, so you can see how each tool's update cadence and visual language changes what the same data communicates.

How it works

• A central model node emits metric pulses that travel along glowing streams to all three stations.
• Matplotlib station only redraws its line chart on a fixed interval — like a static report regenerated on a batch schedule.
• Plotly station updates every frame with a live crosshair on the newest point — like an interactive dashboard an engineer is watching in real time.
• Power BI station aggregates the stream into KPI tiles and a threshold gauge — like an executive summary that only cares whether things are green, amber or red.
• Injecting drift spikes the underlying metric; watch how quickly (or slowly) each station's presentation reveals the problem.

Controls

• Metric stream rate — how often the model pushes a new data point (points/second).
• Matplotlib refresh interval — seconds between static-report redraws.
• Focus audience — highlights the station and camera angle typically used by that audience.
• ⚠ Inject drift / anomaly — pushes a burst of anomalous values into the stream.
• ⟳ Auto-orbit — toggles slow automatic camera rotation. Drag to rotate manually, scroll to zoom.

Did you know?

Many production ML teams run all three of these tools side by side: Matplotlib for reproducible offline reports and papers, Plotly (or Grafana) for live operational dashboards engineers actually debug from, and Power BI or Tableau for the polished, low-refresh summaries that reach non-technical stakeholders.