Harnessing Data Science and AI in Fabric
summary
Shows Fabric running an end-to-end data science flow: created workspace via Fabric API then used a notebook workflow to ingest churn CSV into a Lakehouse, clean with Data Wrangler, do feature engineering and training. Demonstrates MLflow logged three classifiers (two Random Forest configs and LightGBM) with per-run accuracy tracking (~0.86 for one run), model packaging and versioning, batch predict to Delta and Power BI, and roadmap items: real-time scoring endpoint coming plus Copilot integrations (Data Factory, notebooks, Power BI) and OpenAI/SynapseML links and data-boundry notes.
AI-written summary · about
chapters (9)
- 0:00Data science in Fabric
- 5:51Data science workflows
- 14:55Exploratory data analysis
- 17:03MLflow for model training
- 23:58Generating predictions and Power BI
- 26:50Copilot for AI insights
- 31:53Copilot for Data Engineering
- 35:47Copilot for Power BI
- 40:35Recap and Q&A
related talks
22mModels to outcomes with end-to-end data science workflows in Fabric
26mTransform productivity with AI experiences in Microsoft Fabric
19mAccelerate your data potential with Fabric
46mIntegrating Azure AI and Fabric for Next-Gen AI Solutions
50mFabric: What's new and what's next
47mData analytics for the era of AI
27mEmpowering Innovation through DevOps & AI Integration
14mUsing Microsoft Fabric Workload for AuraDB to Make Product Recommendations from OneLake Data