Information Technology Reference
In-Depth Information
putting them in the separate clusters. This is the first step in the proposed method.
C-means clustering technique (explained in Section 3.1) follows the principles of
fuzzy logic, which makes it an appropriate method to deal with the vagueness and
complexity of software modules features. It is able to analyze both numerical and
categorical features, and select the most suitable group for each modules based on
the membership degree. Moreover, it is a robust technique against outlier presence.
Therefore, it is selected to be used for software measurement modules clustering in
the first stage of the proposed method.
The proposed method is organized in two main phases, which are training and
testing stages. In the training stage, the structure of the proposed method is con-
figured and in the testing stage, the software faultiness is predicted. Training and
testing stages are described as follows.
4.1
Training Stage
As in is shown in Fig.1, first 66% of any project is selected for the training purpose.
Then fuzzy clustering is applied on the training dataset and the values of the clus-
ter's centers are recorded. In addition, all produced clusters are also saved along
with their center's weights. We should mention that this experiment is repeated
three times to make sure the different combination of testing and training datasets
are examined. We also trained and tested our proposed model based on selecting
datasets from different available projects in either NASA or Turkish white-goods
manufacturer developing embedded controller datasets [27].
Fig. 1
Training stage of the proposed model
Search WWH ::
Custom Search