الفهرس 
Title : Air pollution forecasting model based on chance theory and intelligent techniques
AbstractOnly 14 pages are availabe for public view
 |  | from | 218 |  |  |
| | | from | 218 | | |
Abstract
This thesis investigated the use of a comprehensive number of machine learning algorithms in the modelling and forecasting the particulate matter air pollutant of diameter less than 10 micrometers (PM10) hourly concentration in the Greater Cairo Metropolitan area (GCMA) in Egypt. Within the literature review conducted in this research, it was found that in the area of Environmental Pollution analysis, the modeling and forecasting techniques, the most popular and widely used learning algorithms are those based on Artificial Neural Networks, Support Vector Machines, etc. from the previous studies, instead of using such single learning algorithms, ensemble learning algorithms were used and were more promising in modeling complex data and have resulted in better prediction accuracies. The purpose of the research conducted in this thesis is to study the modification of a single learning algorithm (SVM), applying the chance weight of the target variable, based on the chance theory, to the corresponding dataset points to be superior than the ensemble learning algorithms. In order to achieve a reliable forecasting model, various algorithms, four monitoring stations, and weighted support vector regression algorithm, have been used in this study. However, the development process passed through seven recursive stages which included variable selection, data collection and preprocessing, segmentation data, support vector regression architecture selection, training process, performance evaluation and the implementation of models using Microsoft Excel, LIBSVM, and WEKA tools. The outcome of this research is a PM10 forecasting model to forecast the PM10 hourly concentration one hour advance in the GCMA in Egypt