Improvement of bagging by increasing probabilistic classifiers’ confidencein prediction: A Case study of SAPCO Parts Supply Company

In today's interconnected global marketplace, the success and competitiveness of businesses areintricately tied to the performance of their supply chains. Supplier selection is a crucial component ofan effective supply chain management, as it directly impacts the supply chain's efficiency, resilience,and overall performance. Choosing capable suppliers is a strategic imperative that can significantlyimpact a company's ability to deliver high-quality products, optimize costs, mitigate risks, and fosterinnovation. Nevertheless, considering various criteria simultaneously, Supplier selection is achallenging process.Studies confirm that Artificial intelligence, particularly machine learning, outperforms traditionalmethods in certain domains due to its ability to handle complex and unstructured data, make accuratepredictions, and adapt to changing conditions. This research, therefore, proposes an improved baggingbasedensemble learning to classify suppliers. In the method, the accuracy of base classifiers is promotedby increasing classifiers' confidence in predicting samples, leading to climbing the accuracy ofensemble learning.The performance of the method was evaluated using a dataset from the supplying Automotive PartsCompany (SAPCO) which is responsible for engineering, design, and parts supply of Iran-Khodro(IKCO), the largest industrial group in Iran. Results represent that the proposed method significantlyenhances the power of bagging-based ensemble learning in evaluating and classifying suppliers.

Keywords:

Ensemble learning , Bagging , DEA Cross efficiency , Supplier selection , Probabilisticclassifier.