h2oai/datatable

Health v2 · maintenance only

Adoption not part of the score

1876 stars · 164 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

datatable is a Python package for manipulating 2-dimensional tabular data structures (data frames), inspired by R's data.table. It emphasizes speed and big-data support (up to ~100GB on a single node) with column-oriented storage and native C++ implementations for all datatypes including strings.

Use cases

• manipulate large csv files faster than pandas
• process 100gb tabular datasets on a single machine
• feature engineering for machine learning pipelines
• read and write csv and jay files quickly
• filter, group, and join large data frames
• replace pandas for memory-efficient data wrangling
• train FTRL-proximal online models on wide datasets

When to choose

• you need faster-than-pandas performance on large tabular data
• you work with datasets too big for pandas memory limits
• you want R data.table-like syntax in Python
• you need native string column handling for ML feature engineering

When to avoid

• you need a rich ecosystem of integrations and community extensions
• your datasets are small and pandas is sufficient
• you rely on pandas-specific APIs or third-party libraries built on pandas
• you need distributed or multi-node processing

Facets

library · maturity maintenance

data-science etl csv math data-science big-data analytics machine-learning python windows dataframe tabular-data high-performance column-oriented big-data pandas-alternative data-table ftrl data-engineering linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="h2oai/datatable")