Leveraging pediatric emergency visits as early signal for respiratory hospitalization forecasting

clustering_code.R - R code that clusters both peaks and finds the cross-correlated cluster to be used for prediction: * Data cleaning (diagnostics and vital sign classification + variable selection) * Hierarchical consensus clustering on the multiple correspondence analysis result * Granger test to define the cross-correlated cluster * Support vector machine to generalize * Verification of the peak cross correlations * Characterization of clusters figure_article.ipynb - Python code that generates daily forecasts for all the models with their results and metrics * Train all the models using one time series * Train all the models using two time series * Figure for the forecasts and the crisis detection * Combine the metrics from all the models and create figures for the metrics diagnostics_R.txt - Document with the list of the diagnostic classification my_dict_results_art.pkl , my_dict_metrics_art.pkl - Dictionary containing the models results and the metrics