• LIGHTNING FAST • POLARS BACKEND • ZERO BLACK-BOX AI • DETERMINISTIC • NO DATA LEAKAGE • AGENTIC ORCHESTRATOR • LIGHTNING FAST • POLARS BACKEND • ZERO BLACK-BOX AI • DETERMINISTIC • NO DATA LEAKAGE • AGENTIC ORCHESTRATOR

API DOCUMENTATION

Welcome to the official documentation for DATADOC. This page details the CLI commands and the Python Core Engine API.

CLI Commands Reference

DATADOC provides a powerful Command Line Interface for rapid dataset engineering.

$ datadoc analyze raw_data.csv --ai
$ datadoc recommend raw_data.csv
$ datadoc engineer raw_data.csv
$ datadoc compare raw_data.csv clean_raw_data.csv
$ datadoc pipeline raw_data.csv
$ datadoc chat raw_data.csv

Python Core API

You can also use DATADOC directly within your Python scripts or Jupyter Notebooks.

from datadoc.core.engine import DATADOC

doc = DATADOC("raw_data.csv")
report = doc.analyze()
clean_df = doc.engineer()
clean_df.write_csv("clean_data.csv")

Core Plugins

The DATADOC engine utilizes 5 deterministic plugins to engineer your dataset:

  • MissingValuePlugin: Imputes missing numeric values with median, categorical with mode.
  • OutlierPlugin: Detects outliers via IQR and clips them dynamically.
  • DatetimePlugin: Extracts temporal features (year, month, day).
  • CategoricalEncoderPlugin: One-Hot Encodes low-cardinality strings.
  • ScalingPlugin: Standard scales numeric columns when max/min variance is high.

AI Chat & Tool Calling

DATADOC incorporates an interactive agentic orchestrator. By running datadoc chat <file>, the AI is granted direct tool-calling access to the underlying Python engine, allowing it to autonomously trigger plugins and clean your data conversationally.