Construction Crew Productivity Prediction By Using Data Mining Methods

Abstract

Ceramic tiling industry has become one of Turkey's fastest growing industries due to the outstanding achievements of Turkish ceramic producers with respect to producing high quality products with lower costs than their equivalents worldwide. Conversely high costs of the end product of Turkish building industry in general show that there is an important problem with the productivity and quality of construction crews. That's why most construction firms begin to realize the need for a detailed research on the factors affecting construction crew productivity. The purpose of this study is thus to classify the factors that affect the productivity of ceramic tiling crews by using data mining methods. To achieve the purpose of our study, a systematic time study was undertaken with ceramic tiling crews in Turkey. Daily productivity values of ceramic tiling crews were collected together with the information related with the factors like the crew size, age and experience of crewmembers. Collected data was classified by using Weka program. The outlier values were first removed from the dataset and decision tree method was used to classify the new dataset. Decision tree method was preferred due to its easiness of use and rapidness in classification. Apriori algorithm, which is the mostly preferred association algorithm in previous studies, was also used to highlight the general trend in the dataset.