Data Workshop Web App

A professional web app for tabular data manipulation with a React UI and Flask API.

Stack

• React (frontend UI)
• Flask (backend API)
• Pandas (data processing)
• NumPy (numeric operations)
• Matplotlib + Seaborn (visual diagnostics)
• Scikit-learn (imputation and scaling workflows)

Features

• Upload multiple input formats: CSV, TSV, TXT, JSON, XLSX
• Remove selected dataset from the workspace with one click
• Navigate dedicated About and Developer Guide pages for onboarding and API reference
• Handle missing values (drop, mean, median, most_frequent, zero)
• Remove duplicate rows
• Normalize column names to snake_case
• Trim whitespace in text columns
• Auto-cast numeric-like text columns using a configurable confidence threshold
• Drop high-missing columns with a configurable threshold
• Configure preview row count (5-100)
• Convert processed output to CSV, JSON, or XLSX before download
• Generate datetime features (year/month/day) from parseable date columns
• One-hot encode categorical columns with a configurable max-level threshold
• Drop constant-value columns
• Clip outliers by sigma bounds
• Scale numeric columns with Min-Max, Standard, or Robust scaling
• Review processing metrics, pipeline step log, and column profile
• Review data quality ratios and an accuracy report (pass/fail/warn checks + score)
• View preview table and statistical summary
• Generate histogram, correlation heatmap, and boxplot diagnostics on demand (optional for faster processing)
• Download the processed file in chosen format

Frontend Notes

• The frontend is built with React and served by Flask.
• React modules are loaded from CDN (esm.sh), so internet access is required when opening the app.
• Processed CSV downloads are served via /api/download/<token> to keep /api/process responses lightweight.
• Additional pages are available at /about and /developer-guide.

Run Locally

python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
python app.py

Open http://localhost:5000 in your browser.