Confused about the different aspects in Machine Learning [closed]

After reading different articles about ML and algorithms, scientist tends to use different words when describing the different aspects in ML.

So now I'm a bit confused myself and I hope you can correct me if I'm wrong.

1) So to my understanding supervised/unsupervised learning are different categories of machine learning algorithms. Each category contains different algorithms such as Neural Networks and Bayesian?

2) Regression, Classification and Clustering are types of models?

3) A model is the result of a trained algorithm?

I hope that I'm not completely wrong, thanks! :)

• Welcome to DS.SE! This is a little bit broad set of questions. Could you make this a single question (=concentrate in one per time)
– mico
Apr 14, 2018 at 15:52
• Thanks! I thought it would be easier to have them in the same question since the answers might overlap with each of the questions. I have updated it with number seperation to make it easier to read/understand :) Apr 14, 2018 at 15:56
• the useability of the site is better when questions are kept separate from each other and each question hits one point, exactly. Apr 14, 2018 at 16:02
• Okay, I will make three different questions instead Apr 14, 2018 at 16:11

Good question and welcome to Datascience

Imagine you have the tree as follows.

                 Machine Learning Models
|
----------------------------------------------------
|                                                  |
Supervised                                         Unsupervised
|                                                  |
- --------------------                                  Clustering
|                    |
Regression          Classification


A model is indeed a trained version of the algorithm you are chosing, regardless of whether its Regression, Classification or Clustering. The above tree is ofcourse a simplified version of all ML algorithms.

• Thank you! :) what would you call supervised/unsupervised learning? Is it just a type of data set structure? Apr 14, 2018 at 20:38
• Say you have, to fit some data to y=ax+b. To do that you are provided multiple individual values of the input (x) and multiple corresponding values of the output (y). Here x would be the input, y would be the output. This is a very simplified version of supervised learning. If on the other hand, if you have a set of inputs (m,n), but the relationship between m, and n is not defined. And you have to find it out using some other algorithm, whether there exists any pattern in the input data set, then you would have to use unsupervised learning. Apr 14, 2018 at 20:51
• quora.com/… Apr 14, 2018 at 20:51
• Short and concise (+1) Apr 15, 2018 at 2:39
• Thank you for your answer! Would you say that neural network is then an algorithm used for fx. classification models? Apr 15, 2018 at 8:08

1) Supervised learning is most of the time the process of learning a mapping, e.g relation, of input features x (sample) to an output y (often labels). Unsupervised learning doesn’t not use labels /output y to learn a relation between the samples and possible labels (ex: clustering).

2) Classification and regression are two types of supervised learning (discrete output labels vs continuous).

Very good resources exist on the forum and web to go deeper with it if you’d like, don’t hesitate.