Approach to Improve the Software Cost Estimation Accuracy

The problems of former software cost estimation methods are pointed out. The fuzzy neural network method is used to construct a quick software cost estimation model and to improve the estimation accuracy. Data of 60 software projects is used as specimen to train the neural network system and fuzzy neural network system. The estimation results of two systems are compared. It shows that the fuzzy neural network model can estimate the software cost more efficient and accurate than neural network model and some former methods.