71. The Drivetrain Approach

To make sure the model you are creating is actually practical, there is method called the Drivetrain Approach introduced by Jeremy Howard.
The steps are the following:

1. Defined Approach- What outcome am I trying to achieve?
2. Levers – What inputs can I control?
3. Data – What inputs can we collect?
4. Models- How does the levers influence our objective?

Using google as an example(which is introduced in the book Deep Learning for Coders with Fastai and PyTorch) the steps above would be something like below.

1. Defined Approach → Show the most relevant search result
2. Levers → The ranking of the search result
3. Data → The implicit information regarding which pages linked to which other pages
4. Models → Changes Traffic