tokyo azure meetup #11 introduction to azure machine learning
TRANSCRIPT
Introduction to Azure Machine Learning Kanio Dimitrov
Use Cases
• Hand written address recognition
• Personal loan approval
• Product recommendation
Machine Learning
Machine learning is a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed. Machine learning focuses on the development of computer programs that can teach themselves to grow and change when exposed to new data.
Major Components
Data Algorithm
Data Analysis Model
• Data drives the prediction logic, not the logic written by developer
Supervised vs Unsupervised • Supervised• Each row has all features• Each row has the value we want to predict
• Unsupervised•We look for clusters of similar data • Algorithm analyzed the data• Algorithm finds similar data
Technique ComparisonSupervised Unsupervised
Value prediction Identify cluster of like dataNeeds training data
containing value being predicted
Data does not contain cluster membership
Trained model predicts value in new data
Model provides access to data by cluster
Machine Learning Workflow
Guidance• Early steps are most important
• Going backwards is typical
• Data always needs “massaging”
• The more data you have, the better
• Never release bad solution with hope to be fixed later
• Analyze results before changing model or dataset
Account OptionsTrial Azure accountNo Azure account required Azure account requiredAuto ML Workspace setup Manual ML Workspace setupAccess only to Experiments and Services
Access to all Azure resources (SQL, Web services mgmt. and more)
Trial account only Permanent accountFree Payment method requiredTrail usage Production usageRetraining not supported Required to retrain model
DEMO – Car Price Prediction
DEMO – German Credit Risk Prediction
Thank you!