Hierarchical Clustering

Disadvantages of Hierarchical Clustering

Work done previously cannot be undone and cannot work well on large datasets.

Types of Hierarchical Clustering

• BIRCH - Balanced Iterative Reducing and Clustering using Hierarchies
• CURE - Clustering Using REpresentatives
• CHAMELEON - Hierarchical Clustering using Dynamic Modeling. This is a graph partitioning approach used in clustering complex structures.
• Probabilistic Hierarchical Clustering
• Generative Clustering Model

Click here to learn Data Science in Bangalore

Density-Based Clustering: DBSCAN

• Clustering based on a local cluster criterion
• Can discover clusters of random shapes and can handle outliers
• Density parameters should be provided for stopping condition

DBSCAN - Density-Based Spatial Clustering of Applications with Noise

Works on the basis of two parameters:

Eps - Maximum Radius of the neighbourhood

MinPts - Minimum number of points in the Eps-neighbourhood of a point

It works on the principle of density

Click here to learn Data Analytics in Bangalore

OPTICS

Ordering of Points to Identify Cluster Structure

Works on the principle of varying density of clusters

2 Aspects for Optics

“Plot the number of clusters for the image if it was subject to Optics clustering”.

Click here to learn Data Analytics in Hyderabad

Grid-Based Clustering Methods

Create a grid structure by dividing the data space into a fixed number of cells.

From the grid's cells, identify clusters.

Challenges:

uneven data distribution is challenging to handle.

is plagued by dimensionality, making it challenging to cluster high-dimensional data.

Click here to learn Artificial Intelligence in Bangalore

Methods:

STING - STatistical INformation Grid approach.

CLIQUE - CLustering in QUEst - This is both density-based as well as grid-based subspace clustering algorithm.

Three broad categories of measurement in clustering:

Click here to learn Artificial Intelligence in Hyderabad

External

Used to compare the clustering output against subject matter expertise (ground truth)

Four criteria for External Methods are:

• Cluster Homogeneity - More the purity, better is the cluster formation.

• Cluster Completeness - Ground truth of objects and cluster assigned objects to belong to the same cluster.

• Ragbag better than Alien - Assigning heterogeneous object is very different from the remaining points of a cluster to a cluster will be penalized more than assigning it into a rag bag/miscellaneous/other category

• Small cluster preservation - Splitting a large cluster into smaller clusters is much better than splitting a small cluster into smaller clusters.

Click here to learn Machine Learning in Hyderabad

Most Common External Measures

• Matching-based measures
  • Purity
  • Maximum Matching
  • F-measure (Precision & Recall)
• Entropy-based measures
  • Entropy of Clustering
  • Entropy of Partitioning
  • Conditional Entropy
  • Mutual Information
  • Normalized Mutual Information (NMI)
• Pairwise measures
  • True Positive
  • False Negative
  • False Positive
  • True Negative
  • Jaccard Coefficient
  • Rand Statistic
  • Fowlkes - Mallow Measure
• Correlation measures
  • Discretized Huber Static
  • Normalized Discretized Huber Static

Internal

Goodness of clustering and an example of same is Silhouette coefficient

Most common internal measures:

• Beta-CV measure
• Normalized Cut
• Modularity
• Relative measure - Silhouette Coefficient

Click here to learn Machine Learning in Bangalore

Relative

Compare the results of clustering obtained by different parameter settings of the same algorithm.

Clustering Assessment Methods

• Spatial Histogram
• Distance Distribution
• Hopkins Statistic

Finding K value in clustering

• Bootstrapping Approach
• Empirical Method
• Elbow Method
• Cross-Validation Method