Case context: You are reviewing a two-step machine learning pipeline for processing images. The system is producing incorrect final predictions, and you need to find the cause without using complex statistical formalisms.

Question: Using the informal method of pipeline error attribution, how should you proceed to diagnose the issue and determine the source of the failure?

Sample answer: You should proceed by inspecting the intermediate output of the first part of the pipeline. By looking at the output of each part, you can evaluate its correctness and decide which specific component made the mistake that caused the final incorrect prediction.

Key points:

• Look at the output of each part of the pipeline.
• Assess the outputs to decide which part made the mistake.
• Use this informal inspection to attribute the error.

Rubric: The answer must describe the process of looking at the outputs of the different pipeline parts to decide which one caused the mistake.