Steve Roe — 45 minutes
This talk describes and demonstrates a Pandas data science workflow that uses the raku Dan and Dan::Pandas modules to apply gradual typing as part of a data grooming exercise between the capture and transformation phases.
Pandas is acknowledged as the leading data analytics package. Types help to detect data inconsistencies and errors early in a data science workflow and to promote consistent patterns for data sharing and reuse. Gradual types and late typing are consistent with the measured application of type control to real-world data.
It shows example code of how to define types at the Series and DataFrame level, and how to apply them to constrain variables and function calls via types and signatures. Examples will also show how to clean data and to coerce types and to resolve type errors. A live demo will be given as a shareable Jupyter console...