Skip to content

ADQA - Autonomous Data Quality Agent

ADQA is a modern, autonomous data quality agent designed to profile datasets, detect multi-dimensional quality risks, and execute automated remediations.

Core Pillars

  • 🔍 Multi-Dimensional Profiling: Deep analysis of structural, behavioral, and semantic characteristics.
  • 🚨 Intelligent Detection: Hybrid engine combining rule-based heuristics with ML-driven risk identification.
  • ⚖️ Quality Scoring: Aggregated scoring across dimensions like Validity, Completeness, Consistency, and Privacy.
  • 🛠️ Autonomous Healing: Automated remediation engine with Advisory, Human-in-the-loop, and Automatic modes.
  • 📜 Full Traceability: Complete audit logs and data lineage for every decision and transformation.

Installation

ADQA is built with Python and leverages poetry for dependency management.

pip install .

For full ML capabilities:

pip install ".[ml]"

Quick Start

Using the Python API

from adqa import ADQA, ADQAConfig, DataSource

# 1. Initialize with a data source
agent = ADQA.from_path("sample_data.csv")

# 2. Run analysis
result = agent.analyze()

# 3. View quality decision
print(result.summary())

# 4. If in automatic mode, remediated data is available
if result.dataframe is not None:
    result.dataframe.to_csv("healed_data.csv")

Using the CLI

adqa analyze sample_data.csv --mode advisory

Documentation Structure

  • Guide: Conceptual overview of how ADQA works.
  • API Reference: Technical documentation of classes and methods.
  • Trace & Lineage: Understanding how decisions are tracked.