Applications and practices
€ - Business tips - €
- ML is not the core, is a tool
- Optimization, no absolute solutions
- Microservices are your friends
Main applications for the web
- Classification
- Regression
- Clustering
- Recommendation
Classification
example: Sentiment analysis
Regression
example: Price estimation
Clustering
example: Personas
Recommendation
example: Collaborative filtering
Issues
- Data: "garbage in, garbage out"
- Model: "nobody is perfect"
Data issues
- Harvest (quantity)
- Cleaning (quality)
- Integration (pertinency)
Model issues
Choosing an algorithm
Reaching 100% accuracy?