M.S. Applied Data Science - Capstone Chronicles 2025

36

Figure 20 Top 20 Most Important Features of Final Random Forest Model Analysis

proportion of actual positives correctly identified by the model, while precision measures the proportion of predicted positives that are truly positive (Kuhn & Johnson, 2013). In the context of product recalls, both metrics are essential: recall ensures that serious risks are not missed, and precision ensures that alerts are trustworthy and not overly alarmist. For Class I recalls—those involving serious or life-threatening health risks—high recall is essential to correctly identify as many true Class I events as possible. A false negative, where a truly

6 Discussion In evaluating the performance of recall severity classification models, it is crucial to assess not only accuracy but also precision and recall. This is especially important for Class I and Class II recalls, which pose the greatest risks to public health. While accuracy measures overall correctness, it fails to reflect how well the model distinguishes between different levels of risk or the consequences of misclassifying critical cases. Recall (also known as sensitivity) measures the

40