Overall major refactoring to support a more streamlined transfer learning approach
Analysis Ready Data: migration to the use of analysis ready data stack, rather than individual s1, s2, dem files
Feature engineering: the process of generating all features (ARD, TTC features, texture) is removed from the deployment pipeline and performed as an independent processing step. Addition of texture_analysis.py
Clean up and consolidation of visualization functions. New functionality to clip final maps to specified shapefile. Addition of quick_viz.py
Training data: exploratory data analysis and cleaning of training data provided by EPA
Notebooks exploring/troubleshooting incorporation of ARD
New functionality to test multi-class regression models in training and deployment
What is the new behavior?
texture_analysis.pyquick_viz.pyDoes this introduce a breaking change?