Abstract:

The present paper presents the results of a research that had as subject the simulation of the US Consumer Credit Fluctuation (CONS) using Artificial Neural Network (ANN). The research objective was to consider the use of ANN as a tool for decision making in credit politics. Initial conditions for the ANN training were a difference between the real data and the simulated data smaller than 1.5%. The researchers used a feed forward artificial neural network and a back propagation algorithm for the training and preparation of future use of the ANN. After the training process the ANN was tested in two sessions, using each time a different sets of data. In order to determine the future ANN use for the CONS forecasting, the research was extended to the simulation of the CONS trend by using the trained ANN and a new set of consecutive values for each of the input data. The trends simulation was a success, when compared the real CONS values with the ANN’s simulated ones the results was a difference with values between [0.000005; 1.20] %. So even with trend simulations the ANN was able to show the training success considering the accuracy smaller than 1.2%. The training and use of ANN was considered a success and the authors considered that the research can be extended to other countries an even to extend the input data by adding others indicators.