Efficient supply chains are critical for the timely delivery of humanitarian supplies and directly influence the effectiveness of public aid initiatives across the world. Accurate prediction of delivery status can be essential for optimising these operations. Thus, this study examines the use of multiple machine learning (ML) algorithms and techniques in order to predict the timeliness of humanitarian deliveries, utilising a comprehensive case-study dataset from USAID, a major international supply organisation. Several ML algorithms were evaluated, including Logistic Regression (aka LR), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Naive Bayes, to assess the predictive model. The analysis yielded weighted recall and accuracy scores between 0.77 and 0.86 across the four algorithms. These results demonstrate the potential of ML methods to forecast delivery status as well as support more proactive and efficient supply chain management (SCM) in global aid contexts. The findings indicate that incorporating advanced predictive analytics into SCM can substantially improve the delivery performance of essential commodities.