Quackalyst the Research Analytics Platform

A platform for training ML models and running statistical analysis
Authors: Luke Rimmo Lego, Samantha Gauthier & Dr. Denver Baptiste

This repository contains a lightweight frontend (React + TypeScript) for loading CSVs, running common statistical tests (t-test, ANOVA, chi-square), training Random Forest models with basic versioning, and visualising results with interactive charts. The UI is purposely small so students can easily understand, modify, and extend it for coursework or research prototypes.

---

Note : This README was written to make it easy for other students and early researchers to download, run, understand, and extend the project. If you carry this forward, please keep changes documented and keep the spirit of reproducibility and accessibility for research tools

---

What you can do with this ?

• Upload CSV datasets and preview them.
• Run descriptive statistics and hypothesis tests.
• Train Random Forest models (client-side) and view basic metrics and feature importances.
• Export model metadata & results as JSON.
• Visualize distributions, scatterplots, boxplots, confusion matrices, and feature importance charts.

---

Quick start : download, install, run locally

Prerequisites

• Node.js (LTS recommended; Node 18+ is a safe baseline).
• npm (bundled with Node) or yarn.
• Git.
• Python 3.9+

Clone & run locally (minimum steps)

# 1) Clone the repo
git clone https://github.com/<your-username>/research-analytics-platform.git
cd research-analytics-platform

# 2) Install JS dependencies
npm ci        # preferred for reproducible installs. We have provided our package-lock.json that was generated
# or
npm install

# 3) Run the dev server
npm run dev

# 4) Open the app in your browser
# by default we use port 8080
# Open: http://localhost:8080

If the app uses Vite/Parcel/webpack, adjust npm run dev command if your project differs.

---

Git & collaboration workflow

If you want to contribute to the project or work in a team, follow this simple GitHub workflow:

# Fork the upstream repo on GitHub (use the Fork button in the web UI)

# Clone your fork
git clone git@github.com:your-username/research-analytics-platform.git
cd research-analytics-platform

# Add upstream to track original repository (so you can sync)
git remote add upstream https://github.com/original-owner/research-analytics-platform.git
git fetch upstream

# Create a feature branch for your work
git checkout -b feat/your-short-description

# Make changes, then
git add .
git commit -m "feat: short description of your change"
git push origin feat/your-short-description

# Then open a Pull Request from your fork to the original repo
# Keep PRs small and focused

Windows users: Git Bash is a friendly shell for these commands. Set core.autocrlf if you see line ending issues:

git config --global core.autocrlf true

---

NPM scripts & common commands

Typical useful commands you will find in package.json:

npm run dev        # start dev server
npm run build      # build for production
npm run serve      # serve production build locally
npm run lint       # run ESLint
npm run format     # run Prettier
npm test           # run tests (if present)

Tip: Use npm ci in CI and reproducible builds; use npm install while developing if you need to add packages.

---

Environment variables & example .env

If you plan to connect to a backend or persist models remotely, create a .env file in the project root with keys like:

# .env.example
REACT_APP_API_URL=http://localhost:5000
REACT_APP_ENABLE_MODEL_PERSISTENCE=false

Copy it and edit:

cp .env.example .env
# open .env in your editor and modify values

---

Roadmap: things future contributors can add

These are concrete tasks that anyone can pick up and implement:

• Add server-side training (FastAPI or Flask) to handle large datasets and async training.
• Add cross-validation and more model types (XGBoost, LightGBM).
• Add hyperparameter search (GridSearch / RandomSearch / Optuna).
• Add user authentication + model registry UI.
• Add dataset lineage tracking via DVC or MLflow.
• Add export to ONNX or PMML for model portability.
• Improve accessibility and mobile responsiveness in the UI.
• Add automated example notebooks (Jupyter) that reproduce experiments end-to-end.

---

Packaging the Application with Electron

This software can be bundled into a desktop application using Electron and electron-builder. This creates a standalone installer (.exe, .dmg, or .AppImage) for Windows, macOS, and Linux.

Prerequisites

Before packaging:

• Ensure all dependencies are installed:

npm install

• Ensure the build is ready:

npm run build

Build and Package

To create a Windows installer using NSIS:

npx electron-builder --win nsis

To create a macOS installer:

npx electron-builder --mac

To create a Linux installer:

npx electron-builder --linux AppImage

Electron builder will generate a release/ folder containing the .exe installer (for Windows) and an unpacked version for testing.

Note

The release/ folder should not be committed. It is for distribution only.

---

Troubleshooting & tips

• If npm run dev shows port conflict, check which process holds the port and kill it, or change the port in vite.config.js.
• Large CSVs in-browser cause memory issues; add server-side ingestion for >50MB datasets.
• If metrics look wrong, confirm correct encoding of categorical labels and that missing values were handled.

Thanks for reading, good luck
Luke Rimmo Lego, Samantha Gauthier & Dr. Denver Baptiste