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
• Classification backend support with confusion matrix generation
• Reproducible run artifacts in runs/

Why It Exists

Many domain experts have high-value data but limited time for coding. ASCENDS provides a guided workflow for:

• selecting inputs and target columns
• training quick baseline models
• understanding model quality through clear metrics and plots

All processing runs locally.

Quick Start

./scripts/install.sh
./scripts/run_gui.sh

Windows (recommended, no PowerShell policy dependency):

scripts\install.bat
scripts\run_gui.bat

Windows PowerShell (optional):

./scripts/install.ps1
./scripts/run_gui.ps1

Open: http://127.0.0.1:7777

Manual path (advanced users):

uv sync
uv run ascends gui

Smoke Test

./test/test.sh

Hyperparameter Tuning (Roadmap)

• Planned next step:
  • expanded tuning + Optuna advanced mode

Portable Bundle (No Setup on Target Machine)

Build on your machine:

bash ./bundle/make_bundle.sh

Windows (recommended):

bundle\make_bundle.bat

Windows PowerShell (optional):

./bundle/make_bundle.ps1

This creates:

• dist/ASCENDS-v<version>-<YYYYMMDD>-<OS>.tar.gz on macOS/Linux
• dist/ASCENDS-v<version>-<YYYYMMDD>-<OS>.zip on Windows

Note: Linux bundles can be significantly larger because XGBoost may pull NVIDIA NCCL runtime wheels. For Linux advanced users, uv sync (or a dedicated conda env) is highly recommended. On the target machine (same OS/arch family), unpack and run:

./launch_gui.sh

Documentation

• Developer and internal details: README.dev.md

License

MIT