| Summary: | 碩士 === 國立臺灣科技大學 === 營建工程系 === 104 === Because there is less information in preliminary design phase, construction industry usually predict project duration by past experience in similar project. Although it can quickly make predictions, accuracy is poor. This research collected several related literatures of project duration, and aggregate Influential factors in preliminary design phase. Further, this research used expert-questionnaire and statistical analysis tools select six Influential factors, collecting forty actual cases based on these factors. Applying Symbiotic Organisms Search-Least Squares Support Vector Machines (SOS-LSSVM) to build Self-Tuned Project Duration Prediction Model. This research assessed model performance by using the K-fold cross validation method. The results showed that the prediction error is less than 10% in MAPE. Compared with other modules, the result is also better than the Regression, Back Propagation Neural Network (BPNN), Support Vector Machine (SVM), Least Squares Support Vector Machines (LS-SVM), Evolutionary Support Vector Machine Inference Model (ESIM)and Evolutionary Least Squares Support Vector Machine (ELSIM). The model of this research can predict more effective and precise.
|
|---|
