Bilgi keşfi amacıyla veri madenciliği algoritmalarının kaya mekaniği deney sonuçları veri ambarı üzerinde uygulanması

Rock mechanics is a fundamental research field of engineering as the mechanical properties of rocks are crucial in mining and civil engineering applications. These properties control main production processes like excavation, drilling, and blasting in addition to geotechnical studies, such as slope stability for surface mining. Experimental studies performed conforming to suggested methods provide essential results representing the mechanical properties of rock material. Within the scope of this thesis study, a database was created containing a total of 9,967 test results, including 284 different projects carried out in the METU Mining Engineering Rock Mechanics Laboratory since the year 2000. After the raw experiment data was prepared by data cleaning operations, it was transferred to the database. OLAP cubes with multidimensional query features were developed to allow advanced analysis by collecting the data in a data warehouse. It is aimed to investigate the potential knowledge discovery of the rock mechanics-related test data by data mining algorithms with the support of the developed data warehouse. A case study was conducted to demonstrate the potential knowledge discovery capability of the data warehouse. In this study, rock types were classified to back fill the missing rock type information using decision tree and random forest algorithms trained. The validation results revealed that the random forest model performed approximately 43 % better than the decision tree model.