Prediction of the project schedule performance outcome with last planner system and social network analysis indicators

Purpose Building earlier last planner system (LPS) only schedule predictors, this study integrates social network analysis (SNA) collaboration metrics and earned value method (EVM) indicators with LPS data to forecast the final schedule performance index (SPI) at planned completion. The approach offers earlier, more actionable diagnostics by having dependence on lagging EVM measures and highlighting team-interaction drivers of schedule risk. Design/methodology/approach The model was developed using machine learning (ML) regression trained on data from 160 projects across ten progress intervals. It incorporated five scope metrics, nine EVM metrics, eight LPS performance metrics and eight LPS collaboration metrics to predict the schedule performance index (SPI) at the end of the initial plan. Data Science and SNA techniques were applied to extract these metrics from LPS software. Findings The LPS-based model using support vector regression achieved an R2 = 0.861 and a mean absolute error of 0.067. The model exhibited greater robustness across 90% of the sample and significantly reduced reliance on result-oriented metrics from 88% to 49%. Research limitations/implications Future research should explore the applicability of the model in different project contexts and further refine the integration of LPS metrics with predictive analytics. Limitations include the reliance on historical data and potential variations in LPS implementation across projects. Practical implications The proposed model provides a proactive approach to schedule compliance prediction, reducing dependence on traditional result-oriented metrics. This enables project managers to anticipate schedule risks earlier and take corrective actions based on LPS performance and collaboration indicators. Originality/value This study presents a novel approach to outcome prediction by leveraging LPS and SNA data for ML regression, demonstrating its potential to enhance schedule performance assessments while shifting the focus from reactive to proactive planning.