Questionnaire data from 520 Thai logistics and supply-chain professionals were clustered into low, moderate and advanced perceived-readiness profiles. SVM best reclassified cluster membership, while SHAP highlighted Actual Use, Technological Factors, Facilitating Conditions and Behavioral Intention.
Key findings
- Three internally derived profiles emerged; SVM achieved the highest accuracy and AUC. Actual use, technology, facilitating conditions and behavioural intention contributed most to profile separation.
Why this matters globally
The clustering-classification-XAI workflow may help emerging-economy logistics sectors tailor support rather than use one-size-fits-all programmes.
Thai researcher contribution
Mahasarakham Business School researchers built a Thai logistics-professional dataset and applied explainable ML to business digital transformation.
Limitations to consider
Responses are perceptions, not audited organisational readiness. Clusters have no ground truth, and classifiers may simply recover the clustering rule; accuracy/AUC are not external predictive validity, and cross-sectional data are non-causal.
Verify the original sources
InformationRead the original article↗DOI: 10.3390/info17070672