Pular para o conteúdo
Kitbay
Skill

data-scientist

NOASSERTIONAtualizado em 11 de nov. de 202670.828Estrelas no GitHubInclui scripts

A skill from the oh-my-openagent project for answering questions about data files such as CSV, Parquet and JSON: grouping, joins, aggregation, time series, distributions, cleaning and charts.

O que a skill ensina

A skill from the oh-my-openagent project for answering questions about data files such as CSV, Parquet and JSON: grouping, joins, aggregation, time series, distributions, cleaning and charts. It picks the cheapest engine that can give a checked answer, preferring a long-running kernel so repeated queries skip start-up and re-reading. After probing file size, free memory and row count, it loads data into memory, queries it in place with DuckDB, streams it with Polars or reads it remotely. Pandas is not allowed, and Excel files must first be exported. Answers include row counts and timings for heavy work, and charts are drawn only when asked or clearly useful. Bundled scripts install dependencies and run quick queries through uv.

O que faz

  • Query CSV, Parquet and JSON files with DuckDB
  • Transform and stream data with Polars
  • Choose in-memory, in-place or streaming by data size
  • Plot trends and distributions with matplotlib