期刊名称：International Journal of Software Engineering and Its Applications
印刷版ISSN：1738-9984
出版年度：2014
卷号：8
期号：2
页码：75-86
DOI：10.14257/ijseia.2014.8.2.09
出版社：SERSC
摘要：This research paper presents a study towards search-based information retrieval and fault prediction with distance functions. The objective of this research is to minimize software costs. Predict the error in software correctly and use the results in future estimation. Through this prediction technique, we have taken different software metrics as input and give output either the software is fault prone or not fault prone or shows any problem in software module in terms of number of errors. This paper presents a work in which we have extended our previous work [12]. In this paper, we discuss an application of Machine learning to error prediction. We have used five different similarity measures namely Euclidean method, Canberra method, Clark method, Exponential method and a Manhattan method to find the best method that increases accuracy. It is observed that the CBR method using the Exponential distance weighted function yielded the best error prediction. In this paper we have used the terms errors and faults, and no explicit distinction made between errors and faults. This software is compiled using Turbo C++ 3.0 and hence it is very compact and standalone, it can be readily deployed on any lower configuration system and it would not impact its performance, as it does not rely on external runtimes and DLL's like the .NET programs rely on. The software is a console based application and thus does not use the GUI functions of the Operating System, which makes it very fast in execution. In order to obtain a result we have used indigenous tool.
