DBSCAN

Density-Based Spatial Clustering Application with Noise(DBSCAN) is a Clustering algorithm using Density-based Method. This algorithm uses density of items to assign items to clusters rather than a centroid or single point.

It requires two parameters:

• : the minimum number of samples that need to be clustered together for an area to be considered high-density.
• : the distance used to determine if a data point is in the same area as other data points.

Advantages:

• DBSCAN is good at handling Outliers and is robust to noise.
• it can create arbitrarily shaped clusters(none-linear and oddly shaped data).
• It can work well even if the shape and number of clusters is unknown.
• It can cluster items with varying densities.

Disadvantages:

• It’s poor at handling lower density data and detecting meaningful clusters, OPTICS is suggested as an alternative.
• It’s poor at identifying clusters of varying density, HDBSCAN is suggested as an alternative.
• it require fine-tuning initial parameters to work well.
• It's not at dealing with high dimensional data.
• It's more complex and computationally expensive compared to simpler models such as K-Means.