Pengukuran Perangkat Lunak Untuk Effort Estimation Dengan Teknik Pembelajaran Mesin


Software effort estimation is to estimate the amount of resources needed in developing the software. For that software effort estimation is important so need to see the effect of software measurement to software effort estimation which is done by machine learning technique. Based on this the researcher tries to build a system capable of measuring software. In this study experiments on software measurement techniques (FPA, FPA with Sugeno fuzzy and FPA with mamdani fuzzy). The three types of techniques are compared with the three project data for further software effort estimation. For evaluation, this study evaluates using the assessment of the Developeras Analyst of the Project. The results of the study that the LOC and effort values on a similar system can be different if calculated by the use of FPA, Fam Mamdany fuzzy and FPA Sugeno Fuzzy. The highest LOC and Effort values are generated by FPA Mamdany Fuzzy on Project DUMAS POLDA SUMSEL. While the lowest effort value and lowest LOC produced by FPA Sugeno Fuzzy. This can be traced from the calculation mechanisms performed by FPA Sugeno Fuzzy where this method does not count the input, output, file, query and interface values at all. The calculation of FPA Sugeno fuzzy is done by roughly judging only from the difficulty of making the system. To raise the price of a project in order to be rewarded higher FAT methods Mamdani Fuzzy is recommended