Sometimes machine learning is not the best method. In this video, explore a set of criteria to determine if machine learning is actually necessary for your use case.
- Machine learning is an exciting topic … and it can be the most exciting project … for a data scientist to be involved in. … But I don't want you to be too eager to apply new techniques … that you take on machine learning projects … that simply won't yield real value. … So I encourage you to assess whether a project … warrants machine learning upfront. … Here's some criteria you can use to determine … if machine learning is actually necessary … and likely to succeed for your use case. … Run through these questions. … Do you have a large and diverse set of data to start with? … If your data source is small, … you're unlikely to achieve meaningful results. … How well defined is the problem you're trying to solve? … Do you have a clear outcome that you were trying to predict … and the hypothesis that you were testing against? … Will a quick ad-hoc analysis suffice … or do you need a full-fledged machine learning model? … Often some straightforward descriptive statistics … could provide the necessary insight. …
This course was created by Madecraft. We are pleased to host this training in our library.