Showing 1 results for Z. Akbari
A. Talebi, Z. Akbari,
Volume 17, Issue 63 (Spring 2013)
Abstract
The real estimation of the volume of sediments carried by rivers in water projects is very important. In fact, achieving the most important ways to calculate sediment discharge has been considered as the objective of the most research projects. Among these methods, the machine learning methods such as decision trees model (that are based on the principles of learning) can be presented. Decision tree method is a hierarchical multi step method which is a recursive data collection technique to binary and smaller sub-divisions until the final analysis cannot be divided. Decision trees consider a priori known set of data and derive a decision tree from it. Then, tree can be used as the set of laws to predict unknown features. In this research, the efficiency of this technique for predicting the suspended sediments in Ilam dam basin has been investigated. To evaluate the accuracy of the methods (written by MATLAB software), statistical criteria such as R, BIAS, RMSE, r2 and MAE were computed. The results showed that based on all the statistical criteria, decision tree in comparison with the sediment rating curve had most consistency with the observed data. Meanwhile, the most important factors for creating tree in the model (that had high correlation with sediment data) are the corresponding discharge and daily rainfall.
