[New page]: addition of Seeding for Spark Function

[donadviser]

Summary of Content

Question: if one does pass seeds into every specific function that requires them, this would avoid any issues with reproducibility or is it still not guaranteed?

Comment: In an ideal world, passing seeds to every function should fix everything. However, in distributed computing, passing seeds is necessary but not always sufficient to guarantee 100% reproducibility.

Language Version

Python

Can this suggestion be used in Pyspark and / or SparklyR (Can select multiple)

Pyspark, SparklyR

Code snippets

Code of Conduct

• I agree to follow this project's Code of Conduct