Ideas¶
Walk-Throughs¶
These are the fully fledged notebooks which will go through all of the preprocessing steps that were done in order to achieve the results we achieved.
- Explore the Data
- Preprocessing - PCA Components Analysis, Column Transformations
- End-to-End Training Example - RF, LR, MLP scikit-learn
- Output Maps (Validation Floats)
- Results
- Line Plots - Variance in Profiles
- Residual Plots
- Joint Distribution Plots
BaseLine Models: sklearn models
- Random Forests
- Linear Regression
- MLP
SOTA Models: Outside
- CatBoost
- XGBoost
Bayesian Models
- Keras:
- Bayesian Layers (Edward)
- Linear Regression (TF Probability)
- Deep Kernel Learning (TF Probability, Bayesian Layers)
GPyTorch Explore
- Exact GP
- Sparse Variational GP
- Deep GP
- DKL - Sparse
Tutorials¶
Feature Extraction
- Column Transformations (scikit-ify everything)
- PCA Trick
CrossValidation
- Model Only
- PreProcessing
Model Training
- Cross Validation (Random, GridSearch)
- Training Tips
- Validation Curves
- Learning Curves
Assessment
- Interpretability
- Permutation Maps
- Sensitivity Analysis
Visualization
- Matplotlib - Complete Customization
- GeoPandas
- XArray
- HoloViews / GeoViews
- DataShader