Deep learning-based forecasting for construction project duration at completion
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Project duration-at-completion (DAC) forecasting is a significant challenge in construction, where inaccuracies can lead to inefficient resource allocation, poor risk management, cost overruns, liquidated damages, and unrealistic stakeholder expectations. Especially during the construction phase, which manages the largest project budget and meets contractual milestones. This research aims to enhance DAC forecast accuracy by leveraging historical data using Deep Learning (DL), providing weekly work packages-level and project-level predictions.A Data Acquisition Model (DAM) collected duration-influencing factors per work package in a time series format, to then apply Long Short-Term Memory (LSTM), One-Dimensional Convolutional Neural Network (CONV-1D), and Multilayer Perceptron (MLP) algorithms. Once the optimal was selected, the overall DAC was computed by consolidating individual predictions and using the current project schedule, Precedence Diagramming and Critical Path Methods. By doing so, LSTM outperformed MLP and CONV-1D, with MASE 0.27, 0.29, and 0.54 for Concrete, Excavation and Backfill work packages. The LSTM-based model surpassed the widely used EVM and ESM, while a Monte Carlo-based sensitivity analysis verified its robustness. This deep-learning model was automated through a GUI, delivering forecasting Gantt charts, performance curves, critical path charts, interacting with Primavera P6 to capture data. This model aims to leverage Artificial Intelligence capabilities in construction.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it