Orange: No-code data mining, visualization and machine learning toolbox (github.com)

Orange is a workflow-driven, no-code data mining and visualization toolbox designed to make exploratory data analysis and basic ML accessible to both novices and experts. The project emphasizes drag-and-drop workflows and extensible widgets so users can inspect, visualize, and model data without writing code, while developers can add domain-specific functionality via add-ons (examples: orange3-text, orange3-bioinformatics, orange3-timeseries, orange3-single-cell, orange3-imageanalytics, orange3-explain). Community channels (Discord) and templates lower the barrier to contributing new widgets. Technically, Orange is distributed as a standalone installer, via conda (recommended), pip, and winget. Typical conda setup: add conda-forge, create a Python 3.12 env, then conda install orange3; add-ons install as conda install orange3-<addon>. If using pip on Windows you’ll need a C/C++ compiler (MS Visual Studio Build Tools). Orange requires PyQt5/PyQt6 and their WebEngine packages and is launched with orange-canvas or python -m Orange.canvas (flags: --help, --clear-widget-settings, --style, -l for logging). Developers should fork the core repositories (orange-widget-base and orange-canvas-core) and install them locally (pip install -e) before orange3 to ensure local changes are used; tests run via Python’s unittest. The release reinforces Orange’s role as a lightweight, extensible platform for teaching, rapid prototyping, and democratizing ML workflows.