Association rule extraction for feature selection, classification and prediction applications and software development

In recent years, together with wide spread use of computer systems, data has been begunto keep in databases and these databases have reached a huge capacity day by day. Therefore,the knowledge discovery from databases has been become an important research area. One ofthe most important methods in this area is Associated Rules Extraction. In this thesis, theassociated rule method is applied on different areas and its performance is evaluated. Asintended for this, three different implementations were carried out:1. Determining relationships between any database quantities, the feature selectionapplication is realized. The obtained results using the feature selection method based onassociated rules are compared with the other feature selection methods.2. The associated rule method is able to extract the relationships between databases. Withthis feature, the texture classification process using the associated rule method is fulfilled.To increase the success rate and speed, the feature extraction process from textures isfulfilled by using the methods of edge detection and wavelet transformation.3. Applying associated rule method on student records, the student scores were analyzed andthe score predictions for the future were carried out in advance. In addition, software tofulfill this purpose was developed.Performing these three applications, it can be seen that the associated rule method canreach an important success rate. After the obtained results were compared with the othermethods in literature, it has been observed that the associated rule method provides anappreciable success rate on the applications of feature extraction, texture classification and scoreforeseeing.