JRuby on Elephants
summary
JRuby glue code calls Apache Mahout for logistic regression and Lucene for tokenizing/stemming to build a kitten detection classifier that trains on ~3,000 labeled tweets (urls stripped to avoid target leakage). The pipeline normalizes, removes stopwords, stems, vectorizes and trains, producing about 93% accuracy on a held-out test set. Notes include practical JRuby gotchas: include jruby-complete jar for stdlib access, use java_class/javasend to disambiguate Java types/overloads, and models can be serialized (large files).
AI-written summary · about
chapters (4)
related talks
46mPractical Machine Learning and Rails
39mNatural Language Processing with Ruby
35mIntelligent Ruby: Getting Started with Machine Learning
27mNatural Language Processing with Ruby
51mMachine Learning with Ruby
46mJRuby on Rails - RoR's Simplicity Meets Java's Class
38mMachine Learning with Clojure and Apache Spark
46mML in Java, YES it's possible!