You are considering building a speech recognition system,

 In the project scoping process, which of these is the first step?

Correct

That's right!

You are considering building a speech recognition system, where the innovation is to use a very low cost microphone that has just been newly invented. Suppose that a human, given an audio clip recorded from this new microphone, often cannot tell what was said (because the microphone has too much noise). What can you conclude from this? 

Correct

There is a high chance it will not be technically feasible for a learning algorithm to achieve high accuracy on this task.

---------

2. You are considering building a product recommendation system. You carried out diligence on feasibility (the ability to give relevant recommendations), but not diligence on value. Which of these is the biggest risk for what could go wrong?

Correct

You build the system and it gives relevant recommendations, but the system does not meaningfully improve key business metrics such as sales conversions or revenue.

Which of these statements is the most accurate?

Correct

There is often a gap between MLE metrics (such as accuracy) and business metrics (such as revenue), and it is useful to try to have the teams compromise and agree on a middle ground that both teams are happy with.

You’ve completed this optional section, when really you didn’t have to. Andrew thinks this means you are (please check all, because all apply):

Correct

Wonderful

You are considering building a product recommendation system. You carried out diligence on feasibility (the ability to give relevant recommendations), but not diligence on value. Which of these is the biggest risk for what could go wrong?

Correct

You build the system and it gives relevant recommendations, but the system does not meaningfully improve key business metrics such as sales conversions or revenue.