Realtime Access Map
Using artificial neural networks to predict issuance durations of occupancy permit applications
This study aims to predict the issuance durations of occupancy permit applications using the delay causes defined in the permit process and reveal the most significant causes affecting the performance of the prediction. Artificial Neural Networks (ANN) is used for predicting the issuance durations of occupancy permit applications. The model is constructed to predict the issuance durations of least once rejected applications made to Izmir Konak Municipality during year 2008. Then, sensitivity analysis is carried out to detect the most significant delay causes affecting the issuance duration. Permit data are examined to reveal the delay causes of occupancy permit process. Six inputs are generated from the delay causes and used in ANN model: 1) Number of missing approval letters, 2) Number of missing payment documents, 3) Number of non-conformances of project to codes and regulations, 4) Number of all missing documents, 5) First permit application season, 6) First permit rejection season. Total issuance durations of the occupancy permit applications are used as the output parameters of the model. The results of the analysis indicate that the prediction accuracy of the model is 86% and the number of missing approval letters, the number of missing payment documents, and the first application season are respectively the three most significant inputs affecting the prediction performance of the model. This study proves that the total issuance durations are so bound to the delay causes in the permit process that it can be learned and predicted by the ANN model and the occupancy permit process is required to be reengineered.