Cash Flow Predictions Using Prophet

Mohd Jaffer Badeghar, Sumedh Mhaske

Abstract: Cash flow forecasting (CFF) is a critical process in the construction industry with significant implications on project viability and financial stability, the efficiency of Artificial Intelligence (AI) in cash flow forecasting for construction projects, comparing the conventional S-Curve method to the AI-based Facebook Prophet model. The research highlighted the growing promise of AI models like neural networks and time-series forecasting methods in improving forecast accuracy, dealing with intricate data, and facilitating proactive decision-making.The approach was to gather cash flow data from an actual residential project in Navi Mumbai. Two models were created: the Prophet model, which is reputed to manage non-linear trends, seasonality, and outliers, and the conventional S-Curve model, which graphs cumulative cash flows against time on the basis of project activity. The Prophet model was implemented with Python libraries Pandas, NumPy, and Matplotlib for data handling and plotting. It was assessed the forecast performance utilizing Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R-squared (R?).The Prophet model had excellent accuracy, Mean Absolute Error (MAE) with a Normalized value of 1.88 %, Root Mean Square Error (RMSE) with a Normalized value of 5.8%, and R-squared (R?) with a value of 0.9996 indicating high precision.In comparison to the S-Curve, it cut down forecasting time by 98% and provided better precision in cash flow forecasting in different construction elements. The model also needed less human intervention and was more responsive to updates of real-time data.The research shows that forecasting through AI, specifically with the application of Prophet, has massive advantages over standard forecasting methods with respect to precision, effectiveness, and time gains.

Keywords: Prophet, Time-Series Forecasting, AI, S-curve