Java in production for Data Mining Research projects
summary
Argues Java is practical for production data‑mining: use Java for heavy ETL (Hive/Pig), fast online services like RTB 5–120 ms targeting, and for model training and deployment via MLlib, H2O, Weka and exporting models as export Java scoring code. Recommends writing memory‑sensitive pieces yourself (for example implement k‑NN yourself) to learn limits; notes Spark helps but typical jobs touch 100 GB Spark apps memory and need GC and caching tuning. Mahout is legacy; pipelines and feature engineering determine success.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:11Java developer in Data Mining
- 10:33Domain and tasks
- 17:55Typical data sources
- 28:01Infrastructure and tools
- 36:44Libraries and frameworks
- 43:26Apache Spark for Java
- 49:24Tips and takeaways
related talks
46mJump-Start Your Data Science Learning Java & Jupyter Notebooks
50mFast Analytics On Big Data
47mScala: The Unpredicted Lingua Franca for Data Science
41mKotlin for Data Science
44mPeople, Technology, Culture: How to make Data Science work
01h 26mData Driven Action: a primer on Data Science
14mDrop Your JAVA and Come Take a Little R! ;)
01h 30mBecome a data-driven Organization with Machine Learning