What is Supervised Learning?
Supervised learning is a machine learning method that maps an input to a desired output. Supervised machine learning algorithms learn by example, like a teacher teaching a class hence the name supervised.
For example if you have say a dataset of 50 pictures of 3 sided figures(triangles) and 4 sided figures(squares), your dataset should include the names of each triangle and square. The supervised machine learning algorithm once trained on this data is then able to figure out the difference between a square and a triangle.
The objective of supervised learning is to predict a correct output for newly presented inputs. In this case if a new image of a square is presented, it should be able to call it a square.
Supervised learning has two categories. These are classification and regression
Classification
A classification algorithm classifies the data into different data points. For example most emails are classified into spam or not spam. A set of parameters are designed into the classification algorithm that allows your actual emails show up in your inbox and the rest in spam. This problem is actually called a binary classification problem.
You can have as many classes as you want depending on the dataset you have access to.
There are a couple of algorithms that can help with classification.
- Linear classifiers
- Support Vector Machines
- Decision Trees
- K-Nearest Neighbour
- Random Forest
Regression
Regression algorithms are used mostly for predictive statistics. The goal is to find relationships between dependent and independent variables. This is usually used to predict a continuous value like in sales, income, sports, studies and so on.
Let’s say we wanted to predict a student’s grade based on how many hours they studied.
This diagram was designed by Aidan Wilson.
From this example we can tell that there is a great relation between the hours being studied and the final test score. The line in blue is called the line of best fit. This shows us how the model would predict a new scenario. If we wanted to know how well a student did after 2 hours of study, the model can tell us the result based on the previous labels it learned from the dataset.
There are three types of algorithms that can help us with the regression problem.
- Linear regression
- Logistic regression
- Polynomial regression
Conclusion
Supervised Learning is usually the introduction to machine learning. It is easy to understand and implement. It is commonly used in many industries and keeps proving to be an excellent tool in use today.
Always Keep Learning
Ediversion 