Loading...

Evaluation of Data Mining in Salinity Prediction and Evaporation Estimation

Mahjoobi, Emad | 2010

772 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 41034 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Agha Mohammad Hossein Tajrishi, Massoud
  7. Abstract:
  8. Since physical variables are almost dependent, uncertain and have significant time and spatial changes, nature of hydrological and environmental systems is so complicated, nonlinear and dynamic. Many models have been developed for studying and analyzing various phenomena in these systems. Recently data mining approaches have been used as new methods for modeling of complicated engineering systems. In this study, performance of several data mining tools in analyzing two phenomena in water quality management have been evaluated. These algorithms are Multilayer Perceptron Neural Network (MLP), Radial Basis Function Neural Network (RBFN), Support Vector Machines (SVMs) in the field of Artificial Neural Networks (ANN), and Classification and Regression Trees (CART) and Model Trees (M5P) in the field of Decision Trees. Ability of these approaches has been studied in the prediction of salinity as an almost linear problem and estimation of pan evaporation as a completely non-linear problem. The data sets used in this research have been extracted from hydrometric and climatologic stations in Karun river watershed. Results indicate the high accuracy and almost similar performance of various techniques in modeling of salinity and evaporation. Decision trees and particularly model trees are more convenient to use, because they are non parametric and able to represent simple and annotative rules. They also determine the important parameters through their algorithms. Besides, they need less run-time and are automatic. Furthermore among artificial neural networks, SVMs were noticed because of almost non parametric approach and lack of over fitting problem and high precision and low run-time. In evaporation estimation, the performance of data mining tools has been compared with empirical models and finally the privilege of data mining tools has been showed by the results
  9. Keywords:
  10. Artificial Neural Network ; Decision Making Tree ; Flamingo ; Evaporation ; Karoun River ; Data Mining

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

No TOC