god — a grammar of data, for Python

One small vocabulary for manipulating tables, spelled the same way in Python, in R, and as plain text. A pipeline is checked whole before any of it runs, so a bad column is reported at the step that names it rather than failing partway through — or worse, not failing.

import pandas as pd
from god import *

sales = pd.read_csv("sales.csv")

sales >> keep(col.region == "West") \
      >> summarize(margin = total(col.margin), by = col.product)

The same sentence runs as plain text, byte-for-byte what you would paste into R or into a database:

run("""
sales
  then keep where [region] is "West"
  then summarize [margin] as total([margin]) by [product]
""")

And when you reach the edge of the vocabulary, it shows you the same pipeline in a tool you already know — show_as(pipeline, "pandas"), or "polars", "pyspark", "dplyr", "sql", "spark".

Installing

pip install grammar-of-data

Then import god. The wheel carries its own engine, so there is nothing else to install.

Every word answers for itself

help(god.keep), help(god.summarize), help(god.add_combinations) — every verb, function and grammar word has a page with a worked example, a small table going in and the answer coming out.

The two names to know

A column is col.name, because Python cannot look one up in your data the way R does. The pipe is >>. Those are the only two differences from the R spelling, and the rest of the vocabulary is identical.

The manual, live in both languages, is at https://psychometrician.github.io/god-book/.