Deep Dive into Azure Automated Machine Learning
summary
Azure Automated ML runs parallel pipelines that do feature selection, hyperparameter tuning and model selection, then returns a deployable model; demo uses a tabular bank marketing CSV, MaxAbsScaler for preprocessing and LightGBM as the top model after about ten minutes of runs, with ROC and confusion-matrix visuals and per-feature importance. Experiments run on an Azure compute instance or Databricks, deploy to a container instance with one click, and provide explainable AI outputs (feature importances, metrics like AUC/F1) for validation.
AI-written summary · about
chapters (24)
- 0:00Start
- 0:42Agenda
- 1:33What Is Machine Learning
- 7:50Selecting the Model
- 8:38Training
- 11:01Model Monitoring
- 12:18What Is Automated Machine Learning
- 18:39When When Do We Need To Use Auto Ml
- 22:03Feature Selection
- 25:22Types of Problems Supported
- 26:57Conflict the Optimal Parameters
- 27:10Why Do I Need To Config the Parameters
- 29:24Imbalance Data
- 30:13Feature Engineering
- 31:08Internal Models
- 32:32Summary
- 33:46Create a Resource
- 33:59Create a Workspace
- 36:02Create a an Auto Ml Run
- 38:26Create Your Own Data Set
- 38:58Configure the Run
- 40:37Model Summary
- 48:44Are There Explanations in Plain Speak for Non Data Science Professionals
- 53:46Free Trial on Azure
related talks
41mLeveraging Automated Machine Learning to enable anyone to develop Machine Learning solutions
49mAutomated Machine Learning on Azure
43mA lap around Azure Machine Learning services
48mStreamlining your Machine Learning projects with Azure Machine Learning
51mTaking Models to the Next Level with Azure Machine Learning
48mMachine Learning in Azure Databricks
01h 05mAzure Machine Learning for the Developer
57mTaking Models to the Next Level with Azure Machine Learning