One of the main missing concepts in a data-driven business is a stable platform for production models. Model factory provides a framework where production models can be easily maintained and monitored. Because the platform was build with a data scientist in mind, it removes the hurdles which commonly occur while moving model from development to production. The platform is build using open source software (e.g., GIT, Jenkins, R) . We explain how to combine these technologies to build your own model factory.
Talk given during the meetup: https://www.meetup.com/The-Amsterdam-Applied-Machine-Learning-Meetup-Group/events/234463593/
3. About KPN
2014 20162015 2017 2018 2019 2020
BM
CM
€ bn
Total TelCo market
1. Revenu declining
2. NPS increasing
3. Cost reduction
4. Simplification
Trend traditional TelCo market: continuous decline
4. About KPN
Data & Analytics
Analytics
Data &
Information
Advanced
Analytics
Customer
interaction
CEO
CCO COO CFO
5. Model development & production before
Data discovery
Test model
Production model
Scheduling
4 months
6. Main components of the model factory before
Orchestrator Compute engine Storage
7. Main components of the model factory now
Orchestrator Compute engine StorageVersion control
8. Model development & production now
Data discovery
Test model
Production model
Scheduling
Data discovery
Test model
Production model
Scheduling
4 months
1,5 months
Before
Now
35. Big picture
Decrease deployment time
Unified model metrics
Unified way of working
More modeling flexibility
Faster response time
Less FTE per model
Better performing models
Model Factory attributes Business value