Software Effort Estimation Using Logarithmic Fuzzy Preference Programming and Least Squares Support Vector Machines

Purpose: Effort Estimation is a process by which one can predict the development time and cost to develop software or product. Many approaches have been tried this probabilistic accurately, but no single technique has consistently successful. There many studies on effort estimation using Fuzzy Machine Learning. For reason, study aims combine Learning get better results.Methods: Various methods combinations carried out in previous research, research tries methods, namely Logarithmic Preference Programming (LFPP) Least Squares Support Vector Machines (LSSVM). LFPP used recalculate driver weights generate Adjustment Point (EAP). The EAP Lines of Code values are then entered as input for LSSVM. output results measured Mean Magnitude Relative Error (MMRE) Root-Mean-Square (RMSE). In study, COCOMO NASA datasets were used.Result: obtained MMRE 0.015019 RMSE 1.703092 dataset, while dataset 0.007324 6.037986. Then 100% prediction meet 1% range actual show that 89,475 5% effort. also level accuracy than Intermediate method.Novelty: This uses combination LSSVM, an improvement from FAHP method different where produces all data training testing, whereas it only small part data.