Use of Model Tree and Gene Expression Programming to Predict the Suspended Sediment Load in Rivers

This paper presents two recent machine learning techniques namely M5 Model Tree (MT) and Gene Expression Programming (GEP) to predict suspended sediment Loads (SSL) in rivers. The MT is a kind of decision tree that has the capability to predict the numeric values with linear regression function at the leaves, whereas GEP is an extension of genetic programming which uses population of individuals and 'survival of the fittest' concept in its evolution, with one or more genetic operators. Both MT and GEP methods are applied for a case study and established relations between SSL and river discharges. To evaluate the performance of developed models, the model results are compared with the results of conventional methods, such as sediment raring curve (SRC) and multiple linear regression (MLR) techniques. The results show that MT gives good performance as compared with the SRC, MLR and GEP models.