ASCENDS

A practical ML toolkit for scientists who want answers, not boilerplate.

ASCENDS (Advanced data SCiEnce toolkit for Non-Data Scientists) helps research users run local machine learning on their CSV data through a GUI and CLI.

What It Does

• Correlation analysis: Pearson, Spearman, MI, dCor
• Model training and prediction workflows
• Regression outputs with parity plots and SHAP feature importance
• Classification backend support with confusion matrix generation
• Reproducible run artifacts in runs/

All processing runs locally — no data leaves your machine.

---

For End Users — Portable Bundle (No Setup Required)

Download the zip for your platform from the Releases page, unzip anywhere, and run:

Platform	Launcher
Windows	Double-click launch_gui.bat
macOS / Linux	./launch_gui.sh

Then open http://127.0.0.1:7777 in your browser.

First launch on Windows may take 1–2 minutes while math libraries compile. Subsequent launches are fast.

No Python, no uv, no installation needed. Everything is bundled.

---

For Developers — Quick Start

Requires Python 3.11+ and uv.

uv sync
uv run ascends gui

Or use the helper scripts:

# macOS / Linux
./scripts/install.sh
./scripts/run_gui.sh

# Windows (cmd)
scripts\install.bat
scripts\run_gui.bat

# Windows (PowerShell)
./scripts/install.ps1
./scripts/run_gui.ps1

Open: http://127.0.0.1:7777

---

CLI Examples

uv run ascends correlation --csv examples/BostonHousing.csv --target medv --task r --view wide
uv run ascends train --csv examples/BostonHousing.csv --target medv --task r --model rf --out runs/boston_rf
uv run ascends parity-plot runs/boston_rf --scope combined --out runs/boston_rf
uv run ascends predict runs/boston_rf --csv examples/BostonHousing_test.csv --out runs/boston_rf/predict

---

Building a Portable Bundle

# macOS / Linux
bash ./bundle/make_bundle.sh

# Windows (cmd)
bundle\make_bundle.bat

# Windows (PowerShell)
./bundle/make_bundle.ps1

Output:

• dist/ASCENDS-v<version>-<YYYYMMDD>-linux.tar.gz
• dist/ASCENDS-v<version>-<YYYYMMDD>-macOS.tar.gz
• dist/ASCENDS-v<version>-<YYYYMMDD>-windows.zip

The Windows bundle includes a full Python distribution — no dependencies needed on the target machine. See docs/windows_bundle_handoff.md for architecture details.

Note: Linux bundles can be significantly larger because XGBoost may pull NVIDIA NCCL runtime wheels. For Linux, uv sync in a dedicated environment is often preferred over the bundle.

---

Smoke Test

# macOS / Linux
./test/test.sh

# Windows (PowerShell)
./test/test.ps1

---

License

MIT