الفهرس يوجد فقط 14 صفحة متاحة للعرض العام
 |  | from | 172 |  |  |
| | | from | 172 | | |
المستخلص
It was observed In the last few years the increasing number of traffic accidents in Egypt. In fact, Egypt is considered as one of the countries most subjected to traffic accidents. The main reasons of traffic accidents. In Egypt may be the human factor (extra speed, sudden stop, the culture of the driver, the lack of knowledge of the traffic rules for the driver, the driver violation of the traffic rules), the vehicle (shortage in the brake, no lights in the vehicle, separation of the tire, explosion of the tire, excess weight or height or protrusion of the cargo), the roads (bad status of the road, dangerous curve In the road, no traffic signs), or the weather conditions (rain or fog).
There is much data about traffic and traffic accidents, which Is distributed and mostly, irregular.
In this thesis, we use data mining In order to Investigate the real reasons of traffic accidents and the relevance of reasons for traffic accidents. We also studied how the factors of the traffic accidents are correlated to each other. This is done by using three data mining techniques: an Association rules technique, a Clustering technique, and a Classification technique. We then used the statistical Correlation technique in order to measure the correlations among different parameters. Finally we compared the results achieved by each data mining technique to the other. Then, we compared the results achieved by the Data Mining techniques to those achieved by the statistical correlation technique in order to conclude the most suitable technique to this data.
The association rules technique produced many results about traffic accidents. The attributes of the accidents were strongly correlated. Concerning the clustering technique, the relationships among the attributes of the accidents were weak and the degree of dissimilarity was
High. Concerning the classification technique, some of the obtained results are in agreement with these obtaIned by the association rules technique.
The statistical technique, correlation coefficient, produced many results that have a weak correlation. Some of the obtained results are In agreement with these obtained by the association rules technique. The statistical technique attributes correlation was less than the association rules technique attributes correlation. In short, the most suitable technique in this study seems to be the association rules technique.
Finally, we suggest a new form for recording traffic accidents, whichh consists of four parts, which are, accident data, drivers’ data, victims’ data, and accident causes. This form should represent the basis of information, which is collected for all accidents.