RINA - Démonstrateur d’utilisation de l’intelligence artificielle pour la gestion opérationnelle des risques naturels gravitaires géologiques | FEREC 2020

Figure 31: heat mapping of the balanced accuracy of bagging tree classifier as the hyper-parameters vary. The combination for each figure are: (a) minimum number of samples per leaf and maximum depth, (b) number of estimators and maximum depth, (c) minimum number of samples per leaf and number of estimators Some general tendencies is that when the min samples leaf and the number of estimators are equal to 1 the models perform worse. It is important to note that when only one estimator is used the model is a simple decision tree as the one on the last section. Lowering the max depth tends to increase the performance of the models. It was decided to use: max_depth=3, min_samples_leaf=9, n_estimators=10. 40