Case context: Your machine learning team is developing a cat classifier and separately tracking its accuracy across four markets: US, China, India, and Other. The product manager notes that the China market has ten times the active user base of the US market, yet the US market generates the highest revenue per user. The team needs to evaluate model iterations using a single-number evaluation metric.

Question: Based on the provided context, decide whether the team should use a simple average or a weighted average to combine these four market metrics, and justify how they should construct the single formula to guide their optimization process.

Sample answer: The team should use a weighted average rather than a simple average to combine the four market metrics. Because the China and US markets have different user volumes and financial values, a simple average would treat all regions equally, leading to suboptimal model selection. By using a weighted average, the team can assign mathematical weights to each region's accuracy metric (for example, weighting by user volume or revenue contributions) to create a single-number evaluation metric that reflects their combined strategic priorities.

Key points:

• Recommend a weighted average over a simple average to combine the four regional metrics.
• Explain that simple averaging fails to account for differences in user volume or revenue value among markets.
• Propose assigning weights to each of the four markets based on business value, traffic, or strategic importance.
• Create a single-number evaluation metric formula that combines all four numbers to guide optimization.

Rubric: The response must recommend a weighted average, explain why a simple average is inadequate given the market disparities, and describe how weights should be assigned based on strategic factors like user volume or revenue.