haku: @author Pivka, M. / yhteensä: 6
viite: 3 / 6
| Tekijä: | Pivka, M. |
| Otsikko: | How many metrics are really necessary? |
| Lehti: | 0
1994 : p. 1369-1375 |
| Asiasana: | |
| Vapaa asiasana: | computer science, software, quality, measurements, programmes, programming languages, COBOL, statistical methods, factor analysis, symposia, Vienna, 1994 |
| Kieli: | eng |
| Tiivistelmä: | One of the major problem with software metrics is the question how many metrics are necessary in praxis to interpret and explain a program code.For this purpose a programm product was developed and measurements on 450 COBOL programs of the real program product were carried out.We wanted to prove or reject the hypothesis that any metric is good enough when this metric is in use during a long period of time and under equal conditions.The hypothesis was verified by factorial analysis.The results of our research show that only with one metric it is impossible to explain a program code i.e. that the metrics complement each other and that their significance is not equal.For this reason the hypothesis was rejected. |
SCIMA