Supervised Learning Algorithms

“high-angle photo of sea waves rushing on shore” by Joseph Barrientos on Unsplash

Just a quick refresher from the previous article — Supervised Learning algorithms are those that are provided with input and labelled output.

The problems that supervised learning algorithms can solve are divided into two broad types –

Source

1. Classification

Classification problems have data whose labelled outputs are in discrete groups / classes (categorical).

For example, the iris data set. This is a very basic data set that contains five columns.

The input columns / features are –

· sepal — length

· sepal — width

· petal — length

· petal — width

The output column / target class is –

· class

The data present in the features’ columns is used to classify these flowers. You can see how this is done in my Jupyter Notebook.

2. Regression

Regression problems have data whose labelled outputs are continuous numbers / values.

For example, the board games data set. This data set consists of several features, such as type, year published, max players, etc. The target column here is the average rating. The average rating can be any number between 0 and 10, hence the values are continuous. The data present in the features’ columns is used to determine what average rating a game will receive. You can see how this is done in my Jupyter Notebook.

Source

The most common and frequently used Supervised Learning algorithms include –

1. Naive Bayes Classifier

2. Linear Regression

3. Logistic Regression

4. K Nearest Neighbors

5. Decision Tree

6. Random Forest

7. Support Vector Machine

8. Linear Discriminant Analysis

Each one will be discussed in future posts, so stay tuned!