Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/saritaphd/hierarchial-clustering

Hiearchial clustering manual with example
https://github.com/saritaphd/hierarchial-clustering

hierarchical-clustering python

Last synced: 3 days ago
JSON representation

Hiearchial clustering manual with example

Awesome Lists containing this project

README 

        
# Hierarchial-Clustering

## 1. Data Preparation:



Begin with a dataset containing your observations (data points) and the features (variables) that describe those observations. Ensure that your data is in a suitable format for clustering, with numeric values or variables that can be converted to numeric.

## 2. Distance Metric:



Choose a distance metric that measures the dissimilarity between data points. Common distance metrics include:

- Euclidean Distance: Measures the straight-line distance between two points in Euclidean space.

- Manhattan Distance: Measures the sum of the absolute differences between corresponding coordinates of two points.

- Cosine Similarity: Measures the cosine of the angle between two non-zero vectors.

Others: You can explore other distance metrics based on your data's characteristics.

## 3. Linkage Method:



The linkage method determines how the distance between clusters is calculated when merging. Common linkage methods include:

- Single Linkage: The distance between two clusters is defined as the minimum distance between any two data points in the clusters. It often results in elongated, chain-like clusters.

- Complete Linkage: The distance between two clusters is defined as the maximum distance between any two data points in the clusters. It tends to create more spherical clusters.

- Average Linkage: The distance is defined as the average of all pairwise distances between data points in the two clusters.

- Ward's Method: Minimizes the variance within clusters and is less sensitive to outliers.

## 4. Dendrogram:



Perform hierarchical clustering using your chosen distance metric and linkage method. This process creates a dendrogram, which is a tree-like diagram showing the hierarchy of clusters. Each leaf represents a data point, and the branches indicate the merging of clusters.

## 5. Dendrogram Interpretation:



Analyze the dendrogram to understand how the clusters are formed. The height at which branches merge corresponds to the dissimilarity between the clusters. Lower branches represent finer-grained clusters, while higher branches indicate broader clusters.

## 6. Cutting the Dendrogram:



Decide how many clusters you want by cutting the dendrogram at an appropriate height. The choice of cut-off depends on the problem and your objectives. You might use the elbow method or other heuristics to make this decision.

## 7. Assigning Data Points to Clusters:



Assign each data point to a cluster based on the cut-off you chose in the dendrogram. The resulting clusters are your final clusters.

Advanced Concepts:



- Handling Large Datasets: For large datasets, hierarchical clustering can be computationally intensive. Use optimized algorithms or techniques like mini-batch hierarchical clustering to handle large datasets efficiently.



- Evaluating Cluster Quality: Employ internal and external validation metrics to assess the quality of your clusters. Common metrics include the silhouette score, Davies-Bouldin index, and adjusted Rand index.



- Hierarchical Clustering in Python or R: Implement hierarchical clustering using libraries such as scipy or scikit-learn in Python, or hclust in R. These libraries offer functions for hierarchical clustering and visualization.



- Agglomerative vs. Divisive Clustering: While agglomerative clustering starts with individual data points and merges them, divisive clustering begins with a single cluster and divides it into smaller clusters. The choice between these approaches depends on the problem.



- Visualization: Create dendrograms, heatmaps, or other visualizations to better understand your clusters and their relationships.



Hierarchical clustering is a flexible and powerful method for grouping data points, and the choice of distance metric and linkage method can significantly impact the results. Experimentation and a deep understanding of your data and problem domain are key to successful hierarchical clustering.