Predicting Percent Plan Complete through Time Series Analysis

AbstractThe percent plan complete (PPC) is a crucial performance metric for a last planner system (LPS). The high positive correlations of the PPC with time and cost performance enable a project team to obtain long-term projections from microachievements. Because predicting PPCs is helpful for project control, this study aimed to investigate the temporal nature of PPCs and develop a time series modeling framework for PPC forecasting based on historical PPCs and the reasons for noncompletion (RNCs). This study found that, although PPCs and RNCs are captured weekly, their impacts on future performance can spread over a longer time span, and future PPCs can be predicted based on historical values. A minimum data time frame of 18 weeks was proposed in the context of the case project. Historical RNCs also impact PPC forecasting. The inclusion of key RNCs can help improve the forecasting accuracy. The findings from this study provide an insight to the hidden temporal nature of the PPC metric resulting from the practical implementation of the LPS. This model can be used as a prediction tool, allowing project teams to anticipate project outcomes and design suitable execution strategies.