The goals of your ML Project
Define the user experience with the prediction system and how they would benefit from it.
How fast or good does the prediction system have to be?
Assessing Impact & Feasibility of your ML project
Impact
What's the business impact?
Who benefit from your model?
Data Availability
How hard it is to acquire data?
How expensive is data labelling?
How much data will be needed?
Accuracy Requirement
How costly are wrong predictions?
How frequently the model needs to be right to be useful?
Problem difficulty
Good publish work on similar projects?
Compute needed for training?
Compute available for deployment?
<aside> 💡 To find high impact ML projects, look for complex parts of your pipeline and places were cheap prediction is valuable
</aside>
<aside> 💰 The cost of ML projects is particularly driven by data availability, but your accuracy requirement also plays a big role
</aside>
What is the value added by your project?