Doing data science with Clojure: the ugly, the sad, the joyful
summary
Argues for a fast exploratory workflow that aims for answers in under two minutes or at most under twenty minutes, using modified Gorilla REPL notebooks with nightly auto-refresh and Hypothesis comments, a custom released-but-undocumented library that treats tables as a vector-of-maps pipeline (no dataframes), composes with transducers and automatic currying, uses Prismatic Schema for early error matching, outsources heavy ML to scikit learn or R, and builds a ggplot-based DSL for better plots while noting editor, interactivity, and optimization gaps.
AI-written summary · about
chapters (18)
- 0:00Intro
- 0:57Big data
- 1:38Timeframe
- 3:20Visualization
- 3:50Audit trail
- 4:40Sharing results
- 6:32Notebook vs wrapper
- 7:20Why not both
- 8:10Notebook
- 11:41Data science tools
- 17:54Compressibility
- 19:56Type systems
- 21:26Interactive environments
- 22:30Machine learning
- 23:37Plotting
- 23:57Charts
- 24:29Wishlist
- 25:40Conclusion
related talks
43mDoing Data Science with Clojure
35m"Clojure for Data Science in the Real World"
18mNavigating Clojure: A Fresh Look From a Data Analysis Background
27mData and Clojure: Using Clojure for more than software development
44mData Science in Clojure
40mClojure Emerging Data Ecosystem
40m"Clojure in the Fintech Ecosystem"
53mCan Haskell Become a Great Language for Data Science?