The system for prediction and optimization in managing cash machines

When attempting to reduce costs in processes such as providing the continuous service of the ATM machines one has to approach a large number of potential problems. These are simple issues such as dealing with the "expert input" but also more demanding issues such as the sheer complexity of the optimization process in terms of the number of variables and the complexity of the algorithm. The aim of our work was, by using Oracle Data Miner and other tools available with Oracle database, to construct a generalizing and easily applicable solution for prediction and optimization of such processes that would lower the cost of maintaining a continuous ATM service. To that end we've developed the algorithm for finding optimal solutions. We've also by using Oracle Data Miner, SQL and PL / SQL solutions created an automated system based on the resulting algorithm for the purpose of minimizing the cost of filling ATMs with cash over a period of time. By further generalization and implementation of features, like for in cases such as the necessity of finding a solution to the traveling salesman problem we've enabled further application of the system on the processes such as management of money distribution for the bank offices. Finally by testing the system it was found that when compared to common practices in banks when dealing with costs of maintaining the ATM service as well when dealing with bank offices our system allows between 5% and 20% reduction in costs. In addition to reducing operational costs, in practice the system ensures that there are no ATMs or bank offices left without available quantity of money at any moment. Finally, the automated application system cancels the need for spending the expensive human resources on the otherwise necessary for optimization, but also extremely time-intensive tasks such as statistical analysis, calculation of the predictions of expenditure, finding the optimal combinations, calculating the global prices and finally finding the optimal solutions.