Author(s): Jadran Berbic; Eva Ocvirk
Linked Author(s): Jadran Berbić
Keywords: Ensemble models; Tree models; Supervised learning; Mean monthly flow; Genetic algorithm
Abstract: Performance of several supervised learning (SL) regression models is compared in order to predict mean monthly flow at hydrological station Vinalic (river Cetina, Croatia). All applied models are tree and/or ensemble models: decision tree regression (DTR), extra trees regression (ETR), random forest regression (RFR), AdaBoost (AB), bagging regression (BR), gradient boosting regressor (GBR), voting regressor (VR) and stacking regressor (SR). Hyperparameters of the models are optimized by genetic algorithm (GA), while their performance is compared on the calibration part of the data.
Year: 2024