So you want to get started with Machine Learning?
There are many machine learning -as a service- platforms which are making the area more accessible as a whole.
You’ve acquired a dataset, picked an environment of choice; but wait! Which of these algorithms should you use?
In this blogpost we'll take a quick look at when we should use which type of algorithm, in layman's terms.
We could divide machine learning algorithms in two ways: by the dataset type and by the algorithm category.
By dataset type
- With labeled data (data which contains labels or tags which include useful information) we use supervised learning
- With unlabeled data (data which does not contain labels or tags) we use unsupervised learning
- When we have a mix of both labeled and unlabeled data we use semi-supervised learning
- When we want to learn our software to be better at something we use reinforcement learning
By Algorithm Category
Then how DO we choose the right algorithm for the job? You may use the diagram below featuring six questions on how to get around this problem.
Using these questions should help with choosing between the various types of algorithms out there. They include the most common scenario's when starting with machine learning but aren't fully exhaustive by any means.