tabular-eda

Installation
SKILL.md

Tabular EDA — Done Right

Whenever you get handed a new tabular dataset, stop. Do not jump straight to XGBClassifier(). Ten minutes of EDA will catch problems that would otherwise destroy your downstream model — target leakage, high-cardinality explosions, MAR missing data, non-linear features that Pearson correlation says are useless. This skill is the workflow.

When to use this skill

  • You just received a new dataset and have no idea what's in it
  • You're about to train a model and want to validate the data first
  • A model is performing suspiciously well (or poorly) and you suspect a data quality issue
Related skills
Installs
5
First Seen
Apr 18, 2026