CA-Markov 土地利用变化预测
CA-Markov land-use change analysis, model training and scenario prediction based on random-forest transition potentials.
CA-Markov land change modelling exposed as three Processing algorithms: transfer
analysis (fast transition-matrix preview), model training (random-forest transition
potential maps) and multi-step scenario prediction with accuracy assessment.
The bundled engine (from the SharpGIS project, MIT licensed) runs in a separate Python
subprocess, so it never loads GDAL into the QGIS process.
Requires numpy, scipy, scikit-learn, rasterio and matplotlib; QGIS 3.28 or newer
(validated on 3.40.15 LTR). All input rasters must share the same grid and CRS.
Outputs: a prediction GeoTIFF, a report, figures and a metrics JSON, shown in a
dockable five-tab result panel.
中文：提供转移分析、训练模型、变化预测三个处理算法，结果以非模态停靠面板展示。