[general]
name=TreeSuit XAI
description=Explainable AI for diagnosing urban-tree growth suitability through spatial validation, diagnosis maps and cell-level Shapley interpretation.
version=2.1.2
qgisMinimumVersion=3.40
qgisMaximumVersion=3.99
author=Yao Chaowen
email=chaowen.yao@aalto.fi
about=TreeSuit XAI is an explainable-AI research workbench for diagnosing urban-tree growth suitability. It combines spatial-block model comparison, locked-test validation, diagnosis rasters, reference-matched environmental Shapley waterfalls and two area-planting alternatives. Requires a separately installed scientific Python with numpy, pandas, scipy, scikit-learn, statsmodels, xgboost, shap, matplotlib, rasterio, joblib and Pillow; see the user guide and requirements.txt. No automatic dependency installation or data upload. The bundled model is Helsinki-specific. Tested on Windows with QGIS 3.40.11; Linux/macOS unverified; QGIS 4 not supported. Experimental screening, not planting approval. User-created joblib model packages must be trusted.
category=Raster
experimental=True
homepage=https://github.com/sleepyheadzzzzzz/Tree-Point-Cloud-Training-and-Analysing
repository=https://github.com/sleepyheadzzzzzz/Tree-Point-Cloud-Training-and-Analysing
tracker=https://github.com/sleepyheadzzzzzz/Tree-Point-Cloud-Training-and-Analysing/issues
icon=icon.svg
license=GPL-3.0
tags=trees,carbon,suitability,explainable AI,spatial validation,shap,planting
changelog=2.1.2: Qt6-scoped Qt and QGIS enums while retaining tested QGIS 3.40 compatibility; predictions unchanged. 2.1.1: public name changed to TreeSuit XAI; always-on raster validation checks, explicit SHA256 digest-byte metadata and code-quality fix; predictions unchanged. 2.1.0: area-planting alternatives, genus counts and polygon exports. 2.0.0: training, locked validation, diagnosis and clickable interpretation.
