Stage 1 samples a Rank Factor Class probabilistically based on an exponential freshness formula \((\frac{1}{1 + n})^{\text{diversity\_weight}}\), where \(n\) is the count of tests previously executed for that class.
flowchart LR
A[Count tests n per class] --> B["Calculate Freshness: (1 / (1 + n))^weight"]
B --> C[Sample Class via Weighted Choice]
C --> D[Stage 2: Bayesian Selection in Chosen Class]