How dexxeth/ml-studio works
Project Type and Purpose
• This is a full-stack machine-learning model management web application.
• Frontend: Next.js (React) UI for dataset upload, preview, feature selection, model training, and download.
• Backend: Python service (main.py) exposing REST endpoints under backend/routes for data ingestion, preprocessing, ML training, and model persistence.
• Database: MongoDB (via backend/database/mongo.py) to store datasets, feature metadata, training run info, and saved model references.
File Structure Analysis
• Top-level directories:
• backend/ – Python service modules (database, preprocessing, models, routes, utils).
• src/app/ – Next.js app pages for each user workflow step.
• src/components/ – Reusable UI components (buttons, tables, cards, alerts).
• src/hooks/, src/lib/, src/styles/ – Client-side utilities, global styles, and custom hooks.
• Key config files: next.config.ts, tsconfig.json, Tailwind (tailwind.config.js), ESLint (eslint.config.mjs), .hintrc.
• ML modules under backend/models/ implement algorithms (SVM, K-means, Random Forest, Linear Regression).
• backend/preprocessing/preprocessor.py cleans and transforms raw datasets.
• backend/utils/ contains helper scripts (clean_nan_inf.py, save_model.py).
README Insights
• Confirms Next.js frontend bootstrapped with create-next-app.
• No explicit backend framework mentioned; infer a lightweight Python HTTP server (Flask or FastAPI) in backend/main.py.
• Workflow: dev server at npm run dev for frontend; likely separate command for Python backend.