Machine Learning is divided by 2 sub-fields: supervised and unsupervised learning, both related to data you use to build a model.
- In supervised learning, the data you use is labelled, i.e. it has a target variable you need to predict, given the response variables . For example, predicting price of an apartment, given such variables as: squared footage, number of rooms, district, area, number of schools nearby, etc.
- In unsupervised learning, the data you use is unlabelled, i.e. it does not have a target variable Y. An example of supervised learning is grouping your customers by segment, based on their characteristics (response variables).
To get a little bit ahead, let’s view on the ML Mindmap, and further dissect it!
Supervised learning is divided into two types of algorithms:
Algorithms that predict a category. Examples can be, an algorithm predicting a movie rating: “Best”, “Good”, “Bad”, “Worst”; an algorithm predicting a fruit: ‘Banana’, ‘Apple’, ‘Orange’; an algorithm predicting any simple yes-no question: “Yes” or “No”, etc.
Algorithms that predict a continuous value, such as “dollars” or “weight”. Examples can be, an algorithm predicting a price for the appartment; an algorithm predicting a weight of a person.
Unsupervised learning is divided into two types of algorithms:
Algorithms that group data based on common characterists that the model would find in the dataset. Examples can be, an algorithm segmenting customers in a market.
Algorithms that generates data, it mostly related to Natural Language Processing, e.g. generating text.