الفهرس | Only 14 pages are availabe for public view |
Abstract Software cost estimation is one of the critical tasks in managing software projects. It is important for software estimation practitioners to understand the full estimation process as described. The importance for this thesis come from the fact that it put a basis for the selection of the estimation methods and techniques in the software estimation domain. The complexity of the environment, and database size affect the estimation accuracy .Artificial neural networks is effective for handling complex environment, while it does not provide the same accuracy if the environment is complex .Also in this thesis, a survey on selection of the data filters gives an overview on how to select the data-filters. The principal conclusions of this research are as follows. When building an estimation prediction model based on historical data. The effect of the data quality on the choosing of the estimation method In our research, we have found that there are methods that affected in terms of estimation accuracy by the data quality more than others: The effect of the Input variable Correlation on choosing of the estimation method. There are estimation methods that are effected if there is a correlation among variables, like linear regression models, while neural networks are not affected with correlation among different variables. The effect of the learning algorithm on the Machine learning algorithms Choosing the machine learning estimation method, is not enough to get an accurate estimation, but also choosing and learning algorithm is also important. |