Year: 2024
Type: article
Abstract: Companies are reconsidering their positions on the data analysis and supply chain changes that have occurred with the onset of AI and its associated features.Supply chain management scholars have attempted a prodigious endeavour in identifying and addressing danger to the supply chain because of ML and DL and their exclusive properties for future data prediction and data classification.To fully tap DL, this study provides several DL approaches towards DL.One of them is to ascertain if the late delivery of a certain product is due to a failure to identify an unforeseen factor in the complex SC system.Thus, the following research questions guide this paper in assessing an effectiveness of Bi-LSTM DL models to enhance a demand forecasting aspect of SC management.The Kaggle dataset titled "Datacom Smart Supply Chain for Big Data Analytics" is used in the SC research.It has been cleaned and scaled before being split into training and testing sets.A BiLSTM model is used to forecast a most important features of late delivery for treatment.Data displaying an F1-Score of 90.11%, AUC-ROC of 14. 01, and an accuracy of 97.59% prove the BiLSTM models are effective.Comparison with other models, such as K-Nearest Neighbours (KNN) and Decision trees, also put into more light the efficiency of BiLSTM in handling SC risks.Overall, the findings suggest that BiLSTM models hold significant promise for optimising logistics operations and mitigating SC risks through more accurate demand forecasting.
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4402660422