Data Science Agent

The AI data scientist that shows its work.

Ask questions about CSV files and databases in natural language. DSA runs SQL, statistics, machine learning, and visualization, then returns claim-level evidence, a reproducible report, and the artifacts needed to inspect how each conclusion was produced.

Try Live Demo · Product Tour · 60-second quickstart · Install from PyPI · Case studies · Evaluation · Docs · v4.3.0

For analysts, researchers, and ML engineers who need answers they can inspect, audit, reproduce, and challenge.

Live browser demo: upload a dataset and run DSA · Static artifact tour: browse verified runs

Why DSA?

Most AI data-analysis tools stop at the answer.

Data Science Agent is built to preserve the path from a claim back to the computation and dataset that produced it.

Question
   ↓
Executed analysis
   ↓
Evidence
   ↓
Claim
   ↓
Reproducible report

That means an analysis can be inspected, audited, reproduced, and challenged instead of accepted on confidence alone.

If an AI makes a data-science claim, you should be able to inspect how it got there.

⚡ 60-second quickstart

Install:

pip install jack-data-science-agent

Run the built-in demo:

dsa demo

Analyze your own data:

dsa analyze sales.csv \
  --task "Which factors explain revenue, and are the effects statistically significant?"

Python 3.12+ is required.

A successful run produces more than chat output:

analysis run
├── report.md
├── experiment.json
├── evidence_graph.json
├── analysis.ipynb
└── reproduce.sh

Hosted browser demo

The verified public demo is available at data-science-agent-web.vercel.app. It runs the Next.js web interface on Vercel and the FastAPI analysis service on Render.

The demo supports the core product flow:

Upload CSV / Excel
      ↓
Inspect dataset
      ↓
Ask a natural-language question
      ↓
Planner selects analysis tools
      ↓
Statistics / ML / visualization execute
      ↓
Evidence is verified
      ↓
Validation + report

A useful smoke-test question is:

Explain which features are most important for revenue, test whether the main associations are statistically significant, assess the impact of campaign_group on the outcome, and clearly distinguish association from causation. Include a visualization.

The hosted preview intentionally runs in deterministic heuristic/offline mode, so it does not require or expose a model API key. The public Render instance uses ephemeral local storage: uploaded datasets and generated artifacts are temporary and may disappear after a restart or redeploy, so re-upload the dataset if it is no longer listed. A free instance may also need a short cold-start period after inactivity.

See Hosted Demo Deployment for the deployment topology, environment variables, and limitations.

Where DSA fits

See it in action

Interactive path: Try the live hosted demo →

Artifact-only path: Open the verified Product Tour →

The Product Tour is a static view of repository artifacts from verified DSA runs. It does not accept uploads or pretend to be a live hosted model/API service. The live demo is separate and is explicitly labeled as a temporary public preview.

_{Question → profile → plan → tools → evidence → critic → reproducible report.}

Start with 3 flagship workflows

Browse all 8 verified case studies →

These are verified repository workflows intended to show product behavior and evidence artifacts. They are not presented as independent real-LLM leaderboard results.

What makes DSA different?

1. Claim-level evidence

Supported findings can be connected to the analysis that produced them:

Claim
└── statistical result
    └── executed tool
        └── parameters
            └── dataset SHA-256

2. Reproducibility by default

DSA records run metadata and reproduction artifacts instead of treating the final prose answer as the only output.

3. Evidence-aware critique

A critic stage checks whether conclusions go beyond available evidence. Causal language can be rejected when the underlying analysis only supports association.

4. Statistical and modeling guardrails

The current analysis path can select semantic outcome/treatment columns, run significance tests, encode categorical features for RandomForest importance, exclude exact target copies from feature importance, and preserve an explicit association-vs-causation boundary.

5. Real data-science tools

Core workflows can coordinate:

• dataset profiling and exploratory analysis
• SQL with DuckDB
• correlation and hypothesis testing
• regression and classification
• forecasting and feature importance
• visualization
• report generation

6. One runtime, multiple interfaces

Python SDK

import asyncio
from data_science_agent import Agent

async def main():
    result = await Agent().analyze(
        "sales.csv",
        "Which region drives the most revenue, and is the trend statistically significant?",
    )

    print(result.report_markdown)

asyncio.run(main())

Inspect evidence programmatically:

for evidence in result.evidence:
    print(evidence.claim)
    print(evidence.source_id)
    print(evidence.result)

Evaluation

DSA is developed against versioned benchmark suites rather than relying only on hand-picked demos.

The repository contains deterministic benchmark and evaluator-validation results, including frozen task suites and reproducibility checks. These results are useful for regression testing and validating the evaluation harness, but they should not be interpreted as an independent comparison of real LLM model quality unless the run identifies a real model/provider and reproducible configuration.

The current public result registry includes a stub/small validation run. It is intentionally labeled as such so test-harness scores are not confused with real-model performance.

• Evaluation methodology
• Benchmark documentation
• Reproducible result registry
• Research & limitations

Architecture

flowchart LR
    A["Natural-language question"] --> B["Planner"]
    B --> C["Data-science tools"]
    C --> D["SQL / Statistics / ML / Visualization"]
    D --> E["Evidence graph"]
    E --> F["Critic"]
    F --> G["Evidence-backed report"]
    G --> H["Reproduction artifacts"]

Hosted demo topology:

flowchart LR
    U["Browser"] --> W["Next.js · Vercel"]
    W --> A["FastAPI · Render"]
    A --> D["Temporary dataset storage"]
    A --> T["DSA tools"]
    T --> R["Evidence + report"]

The runtime uses a LangGraph-based orchestration layer over typed data-science tools, with local-first data operations built around Python, DuckDB, Polars, SciPy, scikit-learn, and Matplotlib.

DSA vs. typical AI data analysis

DSA is not intended to be only another natural-language interface to a dataframe. Its focus is verifiable, reproducible AI data science.

Documentation

Start here:

• Live Demo
• Product Tour
• Getting Started
• Hosted Demo Deployment
• SDK & API Reference
• Evaluation
• Benchmarks
• MCP
• Research & Limitations
• Case Studies
• Roadmap
• Changelog
• Releases

Development

git clone https://github.com/Jackxiaozhiren/data-science-agent.git
cd data-science-agent
uv sync --dev
uv run dsa demo
uv run pytest

Static checks:

uv run ruff check .
uv run mypy .

Contributing

Contributions are welcome, especially:

• real-model benchmark baselines
• reproducibility failures
• new datasets and benchmark tasks
• statistical validation improvements
• data-science tools and plugins
• case studies and documentation

New contributor? Start with the good first issue queue.

See CONTRIBUTING.md, ROADMAP.md, and CODE_OF_CONDUCT.md.

Citation, security, and license

• Academic use: CITATION.cff
• Security reports: SECURITY.md
• License: MIT