Neural nets vs. expert systems: predicting in the financial field

Compares actual data against two methods (a hybrid expert system and a neural network) for predicting a required number based upon past data and known future events. The purpose of the project was to find the best technical approach to predict required values in this type of domain. The significant contributions of this project were applying and comparing AI techniques for prediction of the required loads. The two criteria for success are the accuracy and the reliability of the prediction. Two important results have been obtained: the hybrid expert system predicts better than the human expert, and the neural net has demonstrated that it has the capability at least equal to the expert.<
>