5.0.0
* MAJOR: Milestone release with substantial architecture, performance,
and UX upgrades
* NEW: 12 production-ready ML algorithms with optimization and
explainability workflows
* NEW: Guided wizard + compact dashboard UX, smart defaults, algorithm
comparison, and recipe-driven setup
* NEW: Integrated reporting and diagnostics pipeline for clearer,
reproducible results
* ENHANCED: Faster training/inference paths, stronger validation
workflows, and broader optional dependency support
4.6.0
* NEW: ui/classification_workflow_ui.py - 5-page QWizard with
DataInput, Algorithm, AdvancedOptions, OutputConfig, Review pages
* NEW: ui/comparison_panel.py - AlgorithmComparisonPanel QDialog with
dependency-aware colouring
* NEW: Standalone helper functions (check_dependency_availability,
build_smart_defaults, build_review_summary)
4.5.0
*  NEW: SMOTE oversampling for imbalanced datasets (automatic ratio
detection)
*  NEW: Class weight computation (balanced, uniform, custom
strategies)
*  NEW: Nested cross-validation for unbiased model evaluation
*  NEW: Enhanced metrics (per-class F1, ROC curves, learning curves)
*  NEW: "Nested Cross-Validation" processing algorithm for batch
evaluation
*  NEW: 8 new extraParam keys for imbalance handling and validation
*  ENHANCED: Automatic imbalance ratio detection and strategy
recommendations
*  ENHANCED: Cost-sensitive learning for all supported algorithms
*  ENHANCED: Comprehensive unit and integration tests for imbalance
handling
4.4.0
*  NEW: SHAP explainability - feature importance computation for model
interpretability
*  NEW: "Explain Model (SHAP)" processing algorithm - generate feature
importance rasters
*  NEW: ModelExplainer class with automatic explainer selection
(TreeExplainer/KernelExplainer)
*  NEW: Feature importance raster generation - visualize which bands
matter most
*  NEW: COMPUTE_SHAP, SHAP_OUTPUT, SHAP_SAMPLE_SIZE parameters in
extraParam
*  NEW: scripts/explainability module with comprehensive SHAP
integration
*  PERFORMANCE: Fast TreeExplainer for tree-based models (RF, XGB,
LGB, ET, GBC)
*  PERFORMANCE: KernelExplainer fallback for other models (SVM, KNN,
LR, NB, MLP)
*  ENHANCED: Detailed help strings and usage examples for SHAP
algorithm
*  ENHANCED: Graceful fallback when SHAP unavailable with clear
installation instructions
4.3.0
*  NEW: Optuna hyperparameter optimization - 2-10x faster training
with Bayesian optimization (TPE algorithm)
*  NEW: Factory pattern for classifiers - clean, extensible registry
replacing 700+ line if/elif chains
*  NEW: Custom exception hierarchy with rich context (DataLoadError,
ProjectionMismatchError, etc.)
*  NEW: Clean architecture with separated modules (optimization,
domain, factories)
*  NEW: USE_OPTUNA and OPTUNA_TRIALS parameters (fully backward
compatible)
*  ENHANCED: Comprehensive type hints and docstrings for new modules
*  ENHANCED: Intelligent trial pruning and parallel execution for
Optuna
*  PERFORMANCE: 2-5% accuracy improvement from superior parameter
combinations
*  PERFORMANCE: Random Forest ~3x faster, SVM ~5-8x faster, MLP ~4-6x
faster
*  IMPROVED: Better error messages with actionable suggestions
* FIX: Improved auto-install for Debian/Ubuntu systems where pip
module is not available
* FIX: Auto-install now tries multiple methods: pip module, ensurepip
bootstrap, apt via pkexec
* FIX: Installation failure dialog now shows manual installation
commands directly
* COMPAT: Full QGIS 4.0 forward compatibility - migrated all PyQt5
imports to qgis.PyQt
* PERF: Vectorized scale() functions - ~60% CPU reduction for data
normalization
* PERF: Vectorized Shannon entropy calculation - ~95% speedup (minutes
to seconds)
* PERF: Vectorized raster band reading with GDAL ReadAsArray() - ~30%
I/O improvement
* PERF: Optimized GMM prediction with NumPy einsum - ~40% CPU
reduction
* PERF: Vectorized geodesic distance matrix computation - ~50% speedup
* PERF: Replaced repeated array concatenation with pre-allocation -
~30% memory reduction
* PERF: Replaced if-elif chains with dict lookups for GDAL/NumPy type
conversion
* FIX: Robust layer source path extraction using QgsProviderRegistry
(handles GeoPackage, etc.)
* CODE: Added get_layer_source_path() helper for safe URI parsing
4.2.2
*  FIX: Fixed import errors for splitTrain and trainAlgorithm class
names
*  FIX: Fixed progress_bar.progressBar attribute error (corrected case
to ProgressBar)
*  FIX: Restored missing toolbar icons by fixing Qt resources import
*  FIX: Fixed incorrect resource path in sieve_area.py
*  ENHANCED: Applied ruff linting fixes and improved code quality
*  ENHANCED: Added proper docstrings to sklearn fallback classes
*  ENHANCED: Better exception chaining for improved debugging
4.2.0
*  NEW: 7 additional machine learning algorithms - XGBoost, LightGBM,
Extra Trees, Gradient Boosting, Logistic Regression, Naive Bayes, MLP
*  NEW: Automatic dependency installation system - one-click install
of scikit-learn, XGBoost, LightGBM
*  NEW: Automatic hyperparameter optimization with cross-validation
grid search for all algorithms
*  NEW: Smart sparse label handling - automatically handles missing
class labels (e.g., 0,1,3)
*  NEW: GitHub issue integration - automatic error reporting templates
with system info
*  IMPROVED: Better log levels (INFO vs WARNING) and more informative
progress messages
*  FIX: Resolved parameter delegation issues for XGBoost/LightGBM
wrappers
*  FIX: Fixed model serialization/pickling for new wrapper classes
*  ENHANCED: Real-time pip installation progress with detailed logging
*  ENHANCED: Comprehensive error handling with specific exception
types and user guidance
4.1.0
* Major code refactoring and optimization of
scripts/classification_pipeline.py
* Replaced Hungarian notation prefixes (in/out) with descriptive
parameter names
* Significant memory optimizations for large multi-band image
processing
* Enhanced error handling with specific exception types and detailed
error messages
* Added comprehensive type hints and improved documentation
* Broke down 1279-line method into focused, maintainable helper
methods
* Added configuration constants for better maintainability
* Created parameter migration guide for transition support
* Standardized APIs on descriptive parameter names
4.0.0
* Major version with comprehensive improvements
3.70
* Fix bug with new gdal import from osgeo
3.64
* add closing filter in the processing toolbox
3.63
* fix bug in train algorithm (split waspercent of train not of
validation)
3.62
* fig bug when loading cursor was not removed after unsucessful
learning.
3.61
* fix bug #19 with self.addAlgorithm(alg).
3.6
* Add confidence map in processing
* Add median filter and shannon entropy
* Fix bug with GMM confidence map
* Move dzetsaka icons to extension toolbar