LazyFrame

LazyFrame is a smart lazy DataFrame execution engine with a unified API over multiple backends.

Highlights

• Chainable DataFrame API (filter, select, groupby, agg, sort)
• Logical plan construction with lazy execution (.run() is the trigger)
• Auto backend selection (pandas, polars, duckdb placeholder)
• Simple optimizer with filter pushdown
• Explainability via .explain()
• Actionable hints via .optimize()

Quick start

from lazyframe import lf

df = lf.load({
    "age": [20, 30, 40],
    "country": ["BD", "US", "IN"],
    "income": [1000, 2000, 3000],
})

result = (
    df.filter(df.age > 18)
      .groupby("country")
      .agg({"income": "mean"})
      .sort("income", descending=True)
)

print(result.explain())
print(result.run())

Install for development

pip install -e ".[dev]"
pytest

Backend strategy

• strategy="auto" chooses backend by dataset size heuristic
• strategy="pandas" forces pandas
• strategy="polars" forces polars fallback backend
• strategy="duckdb" forces duckdb placeholder backend