Modeling the organic matter of water using the decision tree coupled with bootstrap aggregated and least-squares boosting

The purpose of the work is to investigate use decision tree (DT) enhanced by bootstrap aggregates (Bag) and least-squares boosting (Lsboost) in modeling organic matter water according its physicochemical parameters. An entire database 500 samples 21 parameters, including matter, was used build DT, DT_Bag, DT_Lsboost models. Training data (364 points) resampled using a technique form different training datasets train models built were validated dataset 91 samples. predicted outputs obtained from developed DT are then combined simple averaging. On other hand, also boosted with Lsboost technical aid increase strength weak learning algorithm. model trains first learner equal weight across all points set, learners based on updated aimed at validation result minimize squared error medium. Good agreement between experimental concentrations for (the correlation coefficient 0.9992), followed DT-Bag 0.9949. comparison revealed superiority mean root errors 0.1295 DT_Lsboost, 0.1664 0.5444 DT). These results show that technology dramatically improved model. This confirmed tests (interpolation 45 points). It should be noted Bag very effective optimizing model, as this close