Test-Driven Data Analysis
summary
Methodology and an open-source Python package tdda deliver concrete results: automatic constraint generation and discover/verify/detect workflows validate data frames (25k account file yields 552 failing cells across 543 records), TDDA serial standardizes messy CSV formats, GenTest auto-generates semantic regression tests, and reference tests + rewritable references enforce reproducible outputs; the library and checklists (22 TDDA-focused) get built and used, catching implementation, applicability and interpretation errors before reports ship.
AI-written summary · about
chapters (10)
- 0:00Test-driven data analysis
- 12:30Why TDDA matters
- 20:53Validating data with constraints
- 35:00Safe CSV handling with TDDA serial
- 43:48Anne and Beth parables
- 53:18Reference testing for analytical code
- 1:05:49Automatic test generation with GenTest
- 1:17:44Comparing data files with TDDA diff
- 1:21:15Test-driven document development
- 1:30:00Q&A
related talks
01h 11mTest-Driven Data Analysis
01h 32mTest-Driven Data Analysis in Python
39mDevelopments in Test-Driven Data Analysis
29mTDD shouldn't be TDDious
15mRobust Coverage with Right Test Data | Need, Challenges, & Benefits
01h 30mAdvanced Software Testing for Data Scientists
24mTDD is not about tests!
41mA Cookbook for Testing Data