Global and subset models workflow

  flowchart TD
    Y0[Dengue data] --> | Filter municipalities \n with 100k inhab| Y[Subset dengue data]
    X1[Weather data]
    X2[Other covariates]
    X1 --> A[Dataset]
    X2 --> A
    Y --> B[Clustering algorithm \n assign k partitions] 
    B --> |Add cluster ID \n variable| A
    A --> A2[Time lag \n outcome and predictors]
    A2 --> |Drop date variable| D{Split}
    D --> E[Train dataset]
    D --> F[Test dataset]
    E --> G[Global model \n specification]
    G --> |Drop \ncluster ID| H[Tune hyperparameters]
    E --> I[Subsets k  \nmodel specification]
    I --> |k models| H
    H --> J[Predict with \n global model training]
    F --> J
    H --> K[Predict with k \n subsets trainings]
    F --> K
    J --> L[Collect metrics for each municipality \n and compare models performance]
    K --> L