Application of stacking ensemble machine learning algorithm in predicting the cost of highway construction projects

Abstract

PurposeThe purpose of this study to apply stacking ensemble machine learning algorithm for predicting the cost of highway construction projects.Design/methodology/approachThe proposed stacking ensemble model was developed by combining three distinct base predictive models automatically and optimally: linear regression, support vector machine and artificial neural network models using gradient boosting algorithm as meta-regressor.FindingsThe findings reveal that the proposed model predicted the final project cost with a very small prediction error value. This implies that the difference between predicted and actual cost was quite small. A comparison of the results of the models revealed that in all performance metrics, the stacking ensemble model outperforms the sole ones. The stacking ensemble cost model produces 86.8, 87.8 and 5.6 percent more accurate results than linear regression, vector machine support, and neural network models, respectively, based on the root mean square error values.Research limitations/implicationsThe study shows how stacking ensemble machine learning algorithm applies to predict the cost of construction projects. The estimators or practitioners can use the new model as an effectual and reliable tool for predicting the cost of Ethiopian highway construction projects at the preliminary stage.Originality/valueThe study provides insight into the machine learning algorithm application in forecasting the cost of future highway construction projects in Ethiopia.

Soft computing in business: exploring current research and outlining future research directions

“…Topic 12- “Forecasting and Predictive Analytics” represents various applications of soft computing-based modelling techniques to make predictions about the future using historical data. The extant literature report several techniques to predict crude oil futures prices (Jiao et al ., 2022; Xu and Niu, 2022), forecasting power load (Dai et al ., 2022), predicting the cost of highway construction projects (Meharie et al ., 2022), energy consumption (Cui et al ., 2022), tourism demand forecasting (Danbatta and Varol, 2022; Li et al ., 2022a, b, c) and stock price prediction (Thesia et al ., 2022).…”

UAV Multispectral Image-Based Urban River Water Quality Monitoring Using Stacked Ensemble Machine Learning Algorithms—A Case Study of the Zhanghe River, China

“…The stacking ML method takes a typical ensemble model based on different learners in which the training data are fed into the first-layer ML model, and the output of the first layer is used as the input of the second layer. This process represents a further search for better approximation based on the first-layer output, which is important for mitigating the risk of overfitting and thus obtaining better prediction results than are realized when using a single ML model [59,60].…”

A Study on the Prediction of House Price Index in First‐Tier Cities in China Based on Heterogeneous Integrated Learning Model

“…In the stacking integration strategy, the selection of the hyperparameters of the base learner effect the overall learning effect and prediction performance of the model, which is a research hotspot and a challenge for the base learner. In existing studies, the selection of hyperparameters for the base learner is usually achieved by utilizing cross-validation [49] and grid search methods [50,51]; however, when there are many parameters or a wide range of parameter values, these methods generate a large amount of computation and reduce the efficiency of the model training [52]. Swarm intelligence is an iterative search algorithm with the following advantages: flexibility, global search, self-organization, and capability for parallel processing.…”