الفهرس Only 14 pages are availabe for public view
 |  | from | 155 |  |  |
| | | from | 155 | | |
Abstract
The present work addresses the problem of the prediction of the climate change, especially El-Nino phenomenon. Data mining techniques are used to study this phenomenon. El-Nino is the anomalous warming of the Sea Surface Temperature (SST) along the equator off the coast of South America. El-Nino has a strong impact on the climate change over the world. Great progress has been made in the effort to study and predict of this phenomenon. High accuracy of climate predictions will lead to significant enhanced economic opportunities, mainly for the national agriculture, fishing, forestry and energy sectors, as well as social benefits.
The thesis presents two techniques based on data mining to predicte this phenomenon. The first one named Auto Regressive Integrated Moving Average (ARIMA). It depends on the model-driven approach. The second one named Artificial Neural Network (ANN) and depends on the data- driven approach.
The Proposed Forecasting System based on Artificial Neural Network (PFSNN) uses back-propagation algorithm in its learning phase. The processing of this system was done through two successive stages. First stage, named PFSNN-A, gets the best architecture of the NN. The second stage includes two methods named PFSNN-B and PFSNN-C. They are adjustments of the output of PFSNN-A through the error modification.
The two datasets monthly and seasonal of the El-Nino3 was applied to the two previous techniques. The procedure addresses the preprocessing of input data, the definition of model architecture and the strategy of the learning process. The main achievement of the thesis is finding out the best model architecture for prediction of El-Nino3.