How to automate the deployment of ...
summary
Automates deploying R models by directly shipping RDS-serialized models to production and calling predict via an embedded R session using the rJava / JRI bridge, avoiding PMML rewrites for unsupported mgcv GAMs. Builds a REST API with PUT/POST/GET for lightweight, low-latency scoring and a Hive generic UDF + DistributedCache workflow for large-batch scoring on Hadoop. Removes training-only data from objects to shrink payloads; REST suits small volumes, UDF on Hadoop scales and proved stable in tests.
AI-written summary · about
chapters (8)
- 0:00Introduction and context
- 3:43Deployment challenges in data science
- 7:02Model as Code approach
- 8:05Development and production setup
- 11:36REST API deployment method
- 15:31Big Data batch deployment
- 26:29Results and perspectives
- 27:14Future improvements and Q&A
related talks
46mProduction Model Deployment
27mDeveloping and Deploying Models at Scale — without Spark!
51mSpeeding up deployment of machine learning models
57mGood Practices for Applied Machine Learning
19mDrive transformation through machine learning
27mDeploying distributed AI and machine learning in financial services
37mBig Data Analytics, Machine Learning, Real Time Processing
53mReal-Time Big Data Analytics with R