Abstract:Purpose The cash flow from government agencies to contractors, called progress payment, is a critical step in public projects. The delays in progress payments significantly affect the project performance of contractors and lead to conflicts between two parties in the Turkish construction industry. Although some previous studies focused on the issues in internal cash flows (e.g. inflows and outflows) of construction companies, the context of cash flows from public agencies to contractors in public projects is still unclear. Therefore, the primary objective of this study is to develop and test diverse machine learning-based predictive models on the progress payment performance of Turkish public agencies and improve the predictive performance of these models with two different optimization algorithms (e.g. first-order and second-order). In addition, this study explored the attributes that make the most significant contribution to predicting the payment performance of Turkish public agencies. Design/methodology/approach In total, project information of 2,319 building projects tendered by the Turkish public agencies was collected. Six different machine learning algorithms were developed and two different optimization methods were applied to achieve the best machine learning (ML) model for Turkish public agencies' cash flow performance in this study. The current research tested the effectiveness of each optimization algorithm for each ML model developed. In addition, the effect size achieved in the ML models was evaluated and ranked for each attribute, so that it is possible to observe which attributes make significant contributions to predicting the cash flow performance of Turkish public agencies. Findings The results show that the attributes "inflation rate" (F5; 11.2%), "consumer price index" (F6; 10.55%) and "total project duration" (T1; 10.9%) are the most significant factors affecting the progress payment performance of government agencies. While decision tree (DT) shows the best performance among ML models before optimization process, the prediction performance of models support vector machine (SVM) and genetic algorithm (GA) has been significantly improved by Broyden–Fletcher–Goldfarb–Shanno (BFGS)-based Quasi-Newton optimization algorithm by 14.3% and 18.65%, respectively, based on accuracy, AUROC (Area Under the Receiver Operating Characteristics) and F1 values. Practical implications The most effective ML model can be used and integrated into proactive systems in real Turkish public construction projects, which provides management of cash flow issues from public agencies to contractors and reduces conflicts between two parties. Originality/value The development and comparison of various predictive ML models on the progress payment performance of Turkish public owners in construction projects will be the first empirical attempt in the body of knowledge. This study has been carried out by using a high number of project information with diverse 27 attributes, which distinguishes this study in the body of knowledge. For the optimization process, a new hyper parameter tuning strategy, the Bayesian technique, was adopted for two different optimization methods. Thus, it is available to find the best predictive model to be integrated into real proactive systems in forecasting the cash flow performance of Turkish public agencies in public works projects. This study will also make novel contributions to the body of knowledge in understanding the key parameters that have a negative impact on the payment progress of public agencies.

Modeling of cash flows from foreign economic activity: Scenario analysis

Simulation modeling as a tool for predicting pedestrian flows when planning a Moscow Metro surface station

Organizing the cash flow management in the construction industry in the Russian Federation

Economic and Mathematical Modeling for the Process Management of the Company's Financial Flows

DEVELOPMENT OF APPLICATION OF THE METHOD OF THE ANALYSIS OF SCENARIOS IN PROFITABLE APPROACH TO ESTIMATION OF COST OF THE OIL COMPANY

Optimization of Financial Flows in a Building Company Using an Escrow Account in the Russian Federation

A data-driven approach for predicting cash flow performance of public owners in building projects: insights from Turkish cases

The Research of Fare Impact on Transport Demand by Mathematical Modeling

Economic and Digital Modeling of Reengineering of Financial Activities of Domestic Enterprises

Innovative Approaches to Improving the Process of Risk Management in the Context of Developing a Strategy for the Foreign Economic Activity of Enterprises

Modeling the Business Environment of an Energy Holding in the Formation of a Financial Strategy

Mathematical Modeling for Financial Analysis of an Enterprise: Motivating of Not Open Innovation

Payments flows system-dynamic model for assessing the efficiency of leasing operations

FINANCIAL SUPPORT AND FORECASTING OF FOOD PRODUCTION USING ECONOMIC DESCRIPTION MODELING METHODS

MODELING CASH FLOWS IN ELECTRONIC PAYMENT SYSTEMS IN REAL TIME

Possibilities of application of ARIMA models in building forecast values of car-flows

Analysis of the economic efficiency of the railway property (Russian experience)

Evaluation of the Effect from Organizational Innovations of a Company with the Use of Differential Cash Flow

FORECASTING INVESTMENT ACTIVITIES OF RAIL TRANSPORT ENTERPRISES USING TREND MODELS

ANALYSIS OF THE MOVEMENT OF CASH FLOWS OF THE ENTERPRISE, THEIR IMPACT ON ECONOMIC ACTIVITY

ASSESSMENT OF RECEIVABLES AS A RISK FACTOR FOR FOREIGN TRADE ACTIVITIES OF CHINESE CHEMICAL ENTERPRISES