Unsupervised Learning
What is the Unsupervised Learning?
If there is no dependent variable in the problem we are interested in, in other words, if there is no label, these types of problems are referred to as ‘unsupervised learning’.
3 Unsupervised Learning Methods
1- K-Means
The aim is to cluster the observations based on their similarities to each other.
How Does K-Means work?
1- Determine the Number of Clusters.
2- Randomly select k centroids.
3- Calculate distances from each observation to the k centroids.
4- Assign each observation to the nearest centroid or cluster.
5- After the assignment process, recalculate centroids for the resulting clusters.
6- Repeat this process for a certain number of iterations, and the clustering structure of observations, where the total within-cluster sum of squares is minimized, is selected as the final clustering.
2- Hierarchical Clustering Analysis
The aim is to partition the observations into subgroups based on their similarities to each other.
What are the differences between K-Means and Hierarchical Clustering?
K-Means pre-determines the number of clusters and iteratively optimizes, while Hierarchical Clustering forms a hierarchical tree structure and allows choosing the number of clusters later.
3- Principal Component Analysis
The fundamental idea is to represent the main characteristics of multivariate data with a fewer number of variables/components. In other words, it is to tolerate a small amount of information loss while reducing the dimensionality of the variables.
If you interested PCA, you can visit PCA Visualization Project in my Kaggle account.
Thank you :)
