Assessing disruptive crises of supply chain in firms amidst COVID-19: An application of multi category-SVM using k-means clustering

ABSTRACT

Supply chains are encountering more uncertain conditions and risks. Disruptions that impede the flow of material through a supply chain that can also result in failure to deliver end goods are a significant category of risks. The consequence of the Covid-19 outbreak has led to shut down production in the supply chain system, resulting in significant impediments for many foreign supply-dependent enterprises. The constraints cause substantial disruptions of the supply chain, production delays, and supplier delays. In recent years, managing supply chain risks has been given more importance to protect supply chains from interruptions by forecasts and prevention. The effects of disruptions on logistics, costs, demand, profits, and inventory levels of the supply chain are analyzed. SVM is one of the most convenient and effective supervised learning algorithms commonly used for classification and regression challenges. This paper presents a modernistic machine learning model, multi-category support vector machines (MC-SVM) algorithm through training on selected samples. In order to abet MC-SVM model to perform well on imbalanced data, k-means clustering algorithm has been proposed to classify clusters of nodes at-disruption, which share similar interruption profiles and can find the relationships between the data object, provide massive information and contribute significantly to accelerating classification and prediction of the SVM model. Data from portfolios of different firms in pharmaceutical industry has been used to train the MC-SVM model which maps the economic performance of a firm to a certain type of supply chain disruption (SCD). The potentiality of this research will privilege better management of the supply chain and thus will permit a network to approach faster response times to the customer, lower costs in all respects of the chain and to the end customer terrific levels of stretch-ability, lower inventories throughout the chain, and diminished the bottleneck effect in the supply chain logistics. © IEOM Society International.