Code and Data to "Quantile regression for temporal streamflow modeling"

This is the accompanying code to "Quantile regression for temporal streamflow modeling", which is part of the manuscript "The Role of Process Heterogeneity in Statistical Modeling", which was submitted to the Austrian Journal of Statistics. The data used in this publication is fully accessible through the LamaH-CE dataset. The two scripts "functions_create_data.R" and "create_data.R" will create the final dataset used for modelling. "functions_modelling.R" provide the functions for tuning the XGBoost model and computing the SHAP values. An example script is also attached (calc_predictions_shap.R). "analyzing_results.R" and "error_metrics.R" will produce the final output used in the manuscript. Finally, two plots produced in the script are added as pdf. All data analysis was performed in R, and we want to acknowledge the following packages: dplyr, tidyr, lubridate, purrr, glmnet, xgboost, shapr, Metrics, gridExtra, zoo and wesanderson.