Exercise: Your next ML project
In this exercise, you should come up with a problem that you would want to see solved by ML, evaluate how well ML could actually solve this problem, and provide first steps towards the solution.
Preferably, this project idea should be related to your current job and may also be a project that you were already planning to work on soon (but no past/completed projects please).
You can use this template (pptx, pdf) to prepare your slides and should aim for a 5 minute presentation.
You might want to review the book section on "Solving problems with ML" again and check out the slides for some example projects below.
In case you need some inspiration for a project idea, have a look at how ML could be used to fight climate change.
Please note: Only well established ML algorithms are listed in the slide template. If your problem requires a different solution, feel free to change one of the answers to a different algorithm category, but you should then probably also add as an additional risk that it might be difficult to solve this problem with ML. If you have trouble determining the correct ML approach for your idea, take another look at the ML algorithm cheat sheet or the book section "ML use cases".
Example project ideas:
[pdf] A manufacturing company wants to improve their production process by automatically recognizing faulty products, understanding why products become defective, and automatically choosing better production settings.
[pdf] A manufacturer of medical imaging machines wants to offer a new feature to automatically highlight problematic areas in the picture that indicate a certain disease.
[pdf] A hardware store wants to extend their online shop with product recommendations.
[pdf] The marketing department of a company wants to better understand their customers so they can improve their marketing campaigns by targeting individual customer groups more selectively.