Web based application maintenance time estimation modeling

The objective of this research was to find out the components and their relationships which effect to Web based transaction processing application maintenance time estimation. Maintenance time indicators were defined from literature review of software maintenance time estimation articles. The 17 maintenance time indicators were used to construct questionnaire. One hundred and forty of public and private sector completed it, software maintenance projects were collected. All of these time indicators which passed significance Factor analysis were then grouped in four Factors (Application Attribute, Application Difficulty, Maintenance Team Attribute, Application Reliability and Modularity). These four Factors were use to find out their relationship with Structural Equation Modeling (SEM). The Factors and Dependence variable (Maintenance time) relationship of Software maintenance time model was adjusted linear relation with generalized additive modeling (GAM) approach using Polynomial modeling fitting. The model was cross validation for accuracy, tested with 30 completed software maintenance projects. The result of Cross validation was presented in Mean Magnitude of Relative Error (MMRE) about 41.36% of model accuracy.