The Right Tools for the Job

• EasyData: A template for Reproducible Data Science
• python3, Cookiecutter, Conda, Jupyter: getting things set up
• Data Science is a DAG: Makefiles and the Data Flow
• Revision Control and git.
  • Yes, you can diff your Jupyter Notebooks
• Automate, Automate, Automate

"Raw data is Read-Only." Sing it with me.

• Fetching and pre-processing data
• Packaging datasets for re-use: the Dataset
• Hashing, numpy, and joblib
• A quick foray into data licenses
• Sharing your datasets: buckets, and DOIs

We're gonna (data) science the *@#! out of this

• The Data Science workflow.
  • EDA and Preprocessing
  • algorithm chains and pipelines
• joblib: caching, hashing, and parallelism
• Running and Recording your Experiments
• Automation: From EDA to scripted data flows

Testing? Already? But I'm not done yet?!

• Types of testing:
  • pytest
  • doctests
  • Hypothesis and property-based testing
• Continuous Integration
• Notebooks as Unit Tests
  • sanitizing problematic text
• Test design patterns

I'm only here for the Pretty Pictures

• Publish or Perish: choices for sharing your data, code, and results.
• Data Visualization in 2018
  • 2d, 3d, interactive
• Source code or it didn't happen.
• Documenting your code
• The crazy world of Software / Data Licenses
• Putting it all together