Welcome to ai4water’s examples!

Notebooks

• Model Building
  • Data Preparation for Regression task
    • defining inputs
    • defining outputs
    • saving prepared data in h5 file
    • Loading from h5 file
    • Multiple Inputs
    • Higher dimensional features
  • Data splitting
    • train fraction
    • val_fraction
    • random splitting
    • reproducibility
    • spliting using indices
    • spliting using intervals
  • Building Machine Learning model for regression
    • Random Forest
      • Training
      • Prediction
      • Evaluation
    • XGBoost
    • Cat Boost
    • Light Graient Boost Machine
    • Custom model/estimator/algorithm
      • uninitiated
      • uninitiated with arguments
      • initiated
  • Data preparation for classification task
  • Building Machine Learning model for classification
    • Random Forest
      • Training
      • Prediction
      • Evaluation
    • XGBoost
    • Cat Boost
    • Light Graient Boost Machine
    • Custom model/estimator/algorithm
      • uninitiated
      • uninitiated with arguments
      • initiated
  • Cross Validation
    • TimeSeriesSplit
    • KFold
    • LeaveOneOut
  • Building Neural Network for regression
  • Building Neural Network for classification
  • Data preparation for time-series prediction
    • Create Data
      • first train batch
      • last train batch
      • first test batch
      • second test batch
      • second last test batch
      • Last test batch
    • Generator using yield
      • Validation data generator
      • Test data generator
  • LSTM for time series prediction
  • Temporal Fusion Transformer Model
  • Loading an existing model from config
• Preprocessing
  • Feature engineering through transformations
  • Missing data imputation
  • HRU discretization
    • unique_sub
    • unique_soil
    • unique_lu
    • unique_slope
    • unique_lu_sub
    • unique_lu_soil
    • unique_lu_slope
    • unique_soil_sub
    • unique_soil_slope
    • unique_slope_sub
    • unique_lu_soil_slope
  • HRU discretization for Laos
• Hyperparameter Optimization (HPO)
  • Hpo for machine learning models (long)
  • Hpo for deep learning models (long)
    • 1) Check model performance with default hyperparameters
    • 2) define objective function
    • 3) define parameter space
    • 4) initial state
    • 5) run optimization algorithm
    • 6) train with best hyperparameters
  • Hpo for machine learning models (short)
  • Hpo for deep learning models (short)
  • Laoding results of hyperparameter optimization
• Experiments
  • Comparing machine learning algorithms for regression
  • Comparing machine learning algorithms for classification
  • Comparing neural network architectures for regression
  • Comparing neural network architectures for classification
  • Comparing performance of RF with data transformations
  • Effect of transformation on LSTM performance
• Postprocessing
  • Analysis of prediction results
  • Partial dependence plot for regression task
  • Partial dependence plot for regression task with categorical features
  • permutation importance for regression task
  • permutation importance with categorical features
  • visualizing layers of neural networks
    • activations as layers
    • As activation functions within layers
  • peeking inside LSTM
    • gradients
    • gradients of activations
    • weights
    • gradients of weights
• Datasets
  • Beach Water Quality of Busan
  • Mtropics dataset from Laos
    • precipitation
    • weather station
    • Water Level and Suspended matter
    • Ecoli
    • pysiochemical attributes
    • rain gauages
    • regression
    • classification
    • ecoli_mekong
  • Global River Water Quality data
    • Pakistan
    • Germany
  • Quadica
    • pet
    • precipitation
    • monthly median values
    • catchment attributes
    • monthly data
    • monthly TN
    • monthly TP
    • monthly TOC
    • monthly DOC

Indices and tables

• Index

• Module Index

• Search Page