Learning-Based Multi-Criteria Decision Making Model for Sawmill Location Problems
arXiv:2604.04996v1 Announce Type: new Abstract: Strategically locating a sawmill is vital for enhancing the efficiency, profitability, and sustainability of timber supply chains. Our study proposes a Learning-Based Multi-Criteria Decision-Making (LB-MCDM) framework that integrates machine learning (ML) with GIS-based spatial location analysis via MCDM. The proposed framework provides a data-driven, unbiased, and replicable approach to assessing site suitability. We demonstrate the utility of the proposed model through a case study in Mississippi (MS). We apply five ML algorithms (Random Forest Classifier, Support Vector Classifier, XGBoost Classifier, Logistic Regression, and K-Nearest Neighbors Classifier) to identify the most suitable sawmill locations in Mississippi. Among these models, the Random Forest Classifier achieved the highest performance. We use the SHAP (SHapley Additive exPlanations) technique to determine the relative importance of each criterion, revealing the Supply-
arXiv:2604.04996v1 Announce Type: new Abstract: Strategically locating a sawmill is vital for enhancing the efficiency, profitability, and sustainability of timber supply chains. Our study proposes a Learning-Based Multi-Criteria Decision-Making (LB-MCDM) framework that integrates machine learning (ML) with GIS-based spatial location analysis via MCDM. The proposed framework provides a data-driven, unbiased, and replicable approach to assessing site suitability. We demonstrate the utility of the proposed model through a case study in Mississippi (MS). We apply five ML algorithms (Random Forest Classifier, Support Vector Classifier, XGBoost Classifier, Logistic Regression, and K-Nearest Neighbors Classifier) to identify the most suitable sawmill locations in Mississippi. Among these models, the Random Forest Classifier achieved the highest performance. We use the SHAP (SHapley Additive exPlanations) technique to determine the relative importance of each criterion, revealing the Supply-Demand Ratio, a composite feature that reflects local market competition dynamics, as the most influential factor, followed by Road, Rail Line and Urban Area Distance. The validation of suitability maps generated by our LB-MCDM model suggests that 10-11% of the MS landscape is highly suitable for sawmill location.
Executive Summary
The article introduces a novel Learning-Based Multi-Criteria Decision-Making (LB-MCDM) framework that synergizes machine learning (ML) with geographic information systems (GIS) to optimize sawmill location selection. By evaluating five ML algorithms, the Random Forest Classifier emerged as the most effective in identifying suitable sites in Mississippi, with SHAP analysis highlighting the Supply-Demand Ratio, proximity to roads/rails, and urban areas as critical decision factors. The study validates that 10-11% of Mississippi’s landscape is highly suitable for sawmill development, offering a data-driven, replicable methodology for timber supply chain optimization. The research underscores the potential of integrating spatial analytics with ML to enhance strategic infrastructure planning in resource-dependent industries.
Key Points
- ▸ Integration of ML and GIS within an MCDM framework for sawmill location optimization, addressing spatial and logistical complexities.
- ▸ Empirical validation in Mississippi demonstrates the Random Forest Classifier’s superior performance among tested ML algorithms, with SHAP analysis quantifying the influence of key criteria.
- ▸ Suitability mapping reveals that only 10-11% of Mississippi’s land is highly suitable for sawmill development, emphasizing the importance of data-driven site selection to avoid suboptimal investments.
Merits
Innovation in Decision-Making Frameworks
The article pioneers a hybrid LB-MCDM approach that combines ML’s predictive power with GIS-based spatial analysis, offering a replicable and unbiased methodology for industrial siting decisions.
Empirical Rigor and Validation
The study employs a robust case study in Mississippi, testing multiple ML algorithms and validating suitability maps, which enhances the credibility and practical applicability of the findings.
Actionable Insights via SHAP Analysis
The use of SHAP values provides transparent, interpretable results by quantifying the relative importance of decision criteria, enabling stakeholders to prioritize factors such as supply-demand dynamics and infrastructure proximity.
Demerits
Geographic and Temporal Limitations
The case study is confined to Mississippi, which may limit the generalizability of the framework to other regions with differing timber supply chains, economic conditions, or regulatory environments.
Data Dependency and Quality Risks
The model’s performance is contingent on the quality and granularity of input data (e.g., GIS layers, supply-demand metrics). Incomplete or outdated data could compromise the reliability of suitability assessments.
Algorithmic Bias and Interpretability
While SHAP improves interpretability, the reliance on ML algorithms (e.g., Random Forest) introduces potential biases in feature selection and weighting, which may not fully capture qualitative or contextual factors (e.g., local political dynamics).
Expert Commentary
This article represents a significant advancement in the application of data-driven methodologies to industrial location problems, a domain traditionally dominated by heuristic or expert-based approaches. The integration of ML with GIS within an MCDM framework addresses a critical gap in the literature by providing a systematic, unbiased, and replicable process for sawmill siting. The empirical validation in Mississippi is particularly noteworthy, as it demonstrates the practical utility of the model in a real-world context. The reliance on SHAP analysis to elucidate feature importance is a commendable step toward enhancing the interpretability of ML-driven decisions, which is often a barrier to adoption in policy or industrial settings. However, the study’s geographic and temporal limitations warrant caution in extrapolating the findings to other regions. Future research could explore the adaptation of this framework to different timber supply chains or incorporate dynamic variables (e.g., climate change impacts on forest growth) to enhance long-term applicability. The article also raises important questions about the scalability of the model for smaller operators or in regions with limited data availability, which could be a barrier to widespread adoption.
Recommendations
- ✓ Expand the geographic scope of the case study to validate the LB-MCDM framework across diverse timber supply chains and regulatory environments, ensuring broader applicability and generalizability.
- ✓ Incorporate dynamic or time-sensitive variables (e.g., timber growth rates, market demand fluctuations) into the model to enhance its adaptability to changing conditions and improve long-term decision-making.
- ✓ Develop an open-source or modular version of the framework to facilitate adoption by smaller operators and policymakers, particularly in regions with limited access to advanced analytics tools or GIS data.
Sources
Original: arXiv - cs.LG