Modeling Delay Percentage of Construction Projects in Egypt Using Statistical-Fuzzy Approach

Abstract

This paper presents two models for predicting the delay percentage in construction projects in Egypt. The first model based on regression analysis. 74 causes that lead to delay in construction projects gathered from literature. A questionnaire survey was made on construction contractors of construction projects in Egypt to evaluate the relative importance of these causes. 14 causes were obtained as the most significant causes that affect the delay percentage (DP) and these are the independent variables of the proposed model. Data for the occurrence of the previous causes on a yes/no basis and the corresponding DP (dependent variable) for 20 construction projects was collected. The data was divided into two sets, the first set contains projects for the purpose of model building. The results revealed that there was a strong linear relationship between DP and 9 causes from 14 causes that significantly affect DP of projects. These causes are: difficulties in obtaining work permits from authorities concerned, original contract duration too short, inflation, difficulties in financing the project by the contractor, effect of subsurface conditions, changes in the scope of the project, economic conditions, excessive bureaucracy in the owner administration, and inefficient coordination by the owner in the early planning &design stages. The second set contains 8 projects for the validation purposes and comparison with the second model. The second model is a statistical fuzzy approach which is a hybrid approach from fuzzy logic and regression analysis. A regression equation between each cause and DP using projects of first set was extracted. The relative weight of each cause is determined by its coefficient of determination (R 2 ) value. The degree of severity each cause had received from questionnaire analysis was used to fuzzify this cause. A trapezoidal membership function was used to represent the delay percentages in construction projects in general depending on 18 out of 30 the previous 20 projects. Two projects were excluded from this function due to their divergence values from other projects. Thus, the expected delay percentage of a project is then determined using fuzzy rules. Validation of the two models using projects of the second set revealed that regression model has prediction capabilities higher than that of statistical fuzzy model. The average percentage error for regression model was 30.3, against 38.5 for statistical fuzzy model.