Shelf-life estimation of fresh produce during transportation using machine learning techniques

Fresh Fruits and Vegetables (FFVs) often suffer significant post-harvest losses due to supply chain inefficiencies, including temperature fluctuations and transit delays. Traditional predictive models, which rely on destructive testing, struggle to adapt to real-world environmental variations. In this work, we present a real-time, non-destructive approach for estimating the Remaining Shelf Life (RSL) of FFVs during transportation. Unlike sensor-based models, our method relies on structured datasets derived from transportation records, incorporating factors such as storage duration, route details, and handling conditions to predict RSL. A machine learning-driven pipeline processes the dataset by handling missing values and encoding categorical variables before training models like Decision Trees and ensemble methods. The proposed framework, tested on real-world logistics data, offers a scalable solution for RSL prediction, enabling proactive quality control and waste reduction across supply chains. Instead of relying on pre-testing, it leverages transport-specific data and environmental interactions to enhance shelf-life estimation.