Machine Learning with Spark and Cassandra: Model Deployment
In Part 6 of our series on Machine Learning with Spark and Cassandra, we will be discussing model deployment. At the end of our machine learning process, we end up with a trained and tested model that we are sure performs to our requirements. Deployment is the process by which that model is made to do the actual processing that it was designed for. Covering:
Model Saving / Transferal / Loading
Service Architecture
Batch vs Realtime Processing
Model Retraining
Previous parts discuss the setup of the environment with Spark and Cassandra, Data pre-processing tactics, and the statistical nature of the tests that underly our validation schemes. We also discussed cross-validation schemes for determining the performance of single machine learning models. Last time we discussed statistical test to use when choosing between two models, even between two different algorithms. Datasets, notebooks, and other assorted code will be made available afterward.
Code for the environment can be found here: https://github.com/HadesArchitect/CaS...
Extra Notebooks and Datasets not included above can be found here: https://github.com/anomnaco/CaSparkEx...
Accompanying Deck: https://www.slideshare.net/AnantCorp/...
Accompanying Blog Post: https://blog.anant.us/machine-learnin...
Awesome Cassandra:
https://github.com/Anant/awesome-cass...
Follow Us and Reach Us At:
Anant:
https://www.anant.us/
Cassandra.Link:
https://cassandra.link/
Email:
[email protected]
LinkedIn:
/ anant
Twitter:
/ anantcorp
Eventbrite:
https://www.eventbrite.com/o/anant-10...
Facebook:
/ anantcorp