Question: In a concise analytical response, connect high bias, training-set fit, and the role of training-data error analysis using only the source concept.

Sample answer: High bias means that the algorithm is not fitting the training set well. In this situation, error analysis on the training data can be useful because it examines the data on which the poor fit is occurring. The analysis may follow a protocol similar to error analysis on the Eyeball dev set and can be performed in addition to other techniques for addressing high bias.

Key points:

• High bias means poor fit to the training set.
• Training-data error analysis can be useful under this condition.
• The protocol can be similar to Eyeball dev set error analysis.
• The analysis can supplement earlier high-bias techniques.

Rubric: A strong response accurately defines high bias through poor training-set fit, explains why training data is the relevant target of analysis, mentions the similar Eyeball-dev-set protocol, and presents the analysis as an additional technique.