Machine Learning is generally divided into two categories
Supervised Learning
UnSupervised Learning
In this article, we are going to talk about Supervised Learning.
Most of the ML Models out there today comes under this category only.
Supervised Learning Algorithm refers to input -> output(/label) mapping
We will provide both input and labels, so that algorithm learns the patterns and come up with the relation between inputs and labels. With that known relation it then tries to give reasonably accurate predictions for the new input data.
Below are some Supervised Learning Examples :
1) Email Spam Classifier
2) Audio to Text Transcript
3) Machine Translation
Mainly there are Two different kinds of algorithms in Supervised Learning
1) Regression
2) Classification
Regression :
A type of supervised learning algorithm, it predicts a number out of infinity many possible outputs. Also called as Continous value prediction.
Example : Housing Prize Prediction
Here in this the possible outcome is one of infinitely possible values like between 100 to 400 thousands of dollars. may be 100.2$ or 125.6$. Can be any number in that range.
Based on the behaviour of data we will build the model that fit to the data may be a straight line in some cases and curve for some cases. So choosing the right model/algorithm is important.
Classification
In Classification, we try to predict out of a small set of possible outcomes. The prediction could be one of a definite set of possible outcomes we know prior.
for example : Using patient's medical records predict whether the tumour is Malignant or Benign.
So here the possible outcomes are only 2, the prediction could be any one of those two only.
So, Classification predicts categories like the picture is Cat or Dog, Tumour is Malignant or Benign, or else Binary classification 0 or 1 lets say if the result is less than 0.5 feedback is Negative or greater than 0.5 result is positive.
For Classification, the algorithms might fit a boundary line. Like for the problem like predicting whether tumour is Malignant or Benign, The algorithm might fit a boundary line that splits both the classes Malignant and Benign.
Thanks for reading! Happy Learning folks!
In the upcoming articles we will learn how regression in particular how Linear Regression works, how we build the model for linear equation and how we fit the weights.
Please feel free to reach out to me on Linkedin :