Unveiling R&D Project Timelines with AI

Predictive analytics in project management has the power to reshape the way we forecast and manage project timelines, paving the way for enhanced efficiency and favourable project results. This research evaluates machine learning models for predicting time required for projects based on project titles and ancillary features. Utilizing feature selection techniques, including Chi-Squared and scikit-learn, we refined our dataset by removing redundant and noisy features to enhance model performance and generalizability. The preprocessing sequence incorporated categorical encoding, and feature scaling to standardize inputs to the model. We implemented and benchmarked predictive models-Random Forest Regressor, Neural Network, Support Vector Regressor, and Gradient Boosting Regressor. Model efficacy was quantified using Mean Squared Error (MSE), Mean Absolute Error (MAE), and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{R}^{2}$</tex> Score. Our results indicate that ensemble methods, particularly Random Forest and Gradient Boosting, exhibit good accuracy in capturing complex nonlinear dependencies within project data. This study demonstrates the significance of meticulous feature engineering and selection in machine learning-driven project management, providing a detailed comparative analysis of model performances for delay prediction. We achieved an MSE of 0.09 through our research, underscoring the effectiveness of our approach in accurately predicting project timelines.