zur Startseite


Analysis Techniques for Claim Data Mining
Betreuer Prof. Dr. rer. nat. Melanie Herschel
Prüfer Prof. Dr. rer. nat. Melanie Herschel
In the framework of this thesis, the claim data from Bosch Rexroth will be analyzed for the feasibility of different business cases such as lifetime prognosis, mean time to failure, etc. Considered scenarios will be realized in the service department by using data mining algorithms. Research of different business cases of lifetime prognosis will be done by comparing the Weibull curves for all Rexroth products and customer’s applications. The application with the shortest Mean Time to Failure (MTTF) in each product family is considered the most critical application. Moreover, researching some other business cases by comparing the prediction with the real number of failures help determining trend warnings and subsequently solving quality problems in an early stage. Another business case is the reduction of downtime in the customer application by doing preventive maintenance. This is supported by conducting predictive analysis and comparing again the prognosis with the real amount of failures. In case there is a good correlation between the two values, it can be assumed that the prognosis will also fit for the next years and the customer can be consulted. In addition, calculating of repair share should be taken under consideration. As a result, the customers with the lowest repair share are identified and can be addressed with tailored service offers. Furthermore, the ratio potential will be identified and in consequence over-dimensioning can be eliminated. Data analysis techniques will be investigated to leverage the data from the existing Bosch Rexroth large claim data in order to predict the characteristics of the product and service such as reliability or probability of failure at a specific time, the mean life and the failure rate. For this data processing a Weibull prognosis analysis will be done in Python.