Figures & data
Figure 1. The research methodology.
![Figure 1. The research methodology.](/cms/asset/2f3ec66d-f5c2-4879-b50a-cbe1e7dce3c0/tprs_a_2262051_f0001_ob.jpg)
Figure 2. A typical cost S-curve of a project.
![Figure 2. A typical cost S-curve of a project.](/cms/asset/291b3e2b-0964-4581-a7e5-607dce0d99b6/tprs_a_2262051_f0002_oc.jpg)
Table 1. The EVM data of a hypothetical project (the tracking period – month 3).
Table 2. The EVM data and inputs to the XGBoost model (the early-stage estimation).
Figure 3. The training-testing protocol of the ML algorithm.
![Figure 3. The training-testing protocol of the ML algorithm.](/cms/asset/f9d26989-e369-48bf-97f9-1b0a6436ae0c/tprs_a_2262051_f0003_ob.jpg)
Table 3. The ML models, their parameter settings and explanation.
Table 4. The accuracy results of the forecasting models, MAPE%.
Table 5. The accuracy results of the forecasting models, NRMSE.
Figure 4. The number of more accurate cases (frequency of projects) by the XGBoost and index-based benchmark models.
![Figure 4. The number of more accurate cases (frequency of projects) by the XGBoost and index-based benchmark models.](/cms/asset/2fe89c6d-9e21-4170-9da9-f60d46c8da31/tprs_a_2262051_f0004_ob.jpg)
Data availability statement
The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.