An Intelligent Hybrid Model for Determining Public-Private Partnership in Iranian Water and Wastewater Industry Based on Collective Tree Algorithms

One of the pillars of any country’s development is access to safe water and sanitation, so it is important to execute water and wastewater projects in the shortest possible time. In this regard, considering the emergence of various methods of partnership, choosing the right approach has become one of the most important issues in this industry. Therefore, a proper investment method in this field has always been the concern of decision makers. Using the database of partnership projects and data mining algorithms in the water and wastewater sector, we have designed a model to predict a proper way for public-private partnership projects. In this research, CRISP data mining method was applied to the data from 176 projects. After understanding and identifying the data, they were cleaned by deleting outliers and noisy data, and missing values were replaced. Then, the process of data classification was performed using decision tree and machine learning algorithms, and necessary analysis was performed. Also, the indicators of PPP were extracted and prioritized. K-fold cross validation method is used for validation and dividing the data. Based on the modeling and considering the data preparations and data mining methods, the stacking method is suitable for predicting and determining the type of public-private partnership contract in the implementation of each project of water and wastewater industry, which has an accuracy of 86.27%. In the pre-processing section, the combined COF method for deleting outliers and entropy factors for feature selection was used. Using the model, the success rate of each project can be predicted and an appropriate PPP contractual template for any water and wastewater project can be proposed. In addition, by entering the information of each new project, the impact of the improvement of each indicator can be easily examined.