K- Means Clusters Example

K- Means Clusters Example

Problem Statement

We have collected data on several different models of cars with respect to several parameters. Use a Kmeans analysis to cluster the different vehicles together.

How to perform analysis

Step 1: Open Sigma Magic
  1. Click on the Sigma Magic button on the Excel toolbar.
  2. Click on the New button to create a new project.
Step 2: Add the analysis template
  1. Click on the Tool Wizard to add the analysis template.
  2. Click on Analytics and then on K-Means Clusters.


Step 3: Specify analysis options
A new worksheet will be added to your workbook. Analysis Setup will be automatically opened, in the setup tab specify the survey results.
​

Click on Data to specify the data required for this analysis.



Click the Verify tab to ensure all the inputs are okay and shown in a green checkmark.



Step 4: Generate analysis result
Click OK and then click Compute Outputs to get the final results.




Interpretation of Results

1.  Cluster Formation:
  • The K-Means algorithm identified 3 clusters in the dataset.
  • Cluster sizes are 7, 9, and 16 data points respectively, indicating an uneven distribution.

2.  Distance Metric & Algorithm:

  • The clustering is based on Euclidean distance, which measures similarity between points.
  • The Hartigan-Wong algorithm (Auto-selected) was used for clustering.

3.  Cluster Visualization:

  • The cluster plot shows well-separated groups, suggesting a reasonable clustering structure.
  • Overlapping regions indicate some similarity between certain data points in different clusters.

4.  Within and Between Cluster Sum of Squares (SSQ):

  • Within-cluster SSQ (sum of squared distances within each cluster) values are 11816.35, 46582.69, and 32462.91.
  • Between-cluster SSQ is 53139.2, meaning a significant portion of variance is due to differences between clusters.
  • A lower within-cluster SSQ and higher between-cluster SSQ indicate well-separated clusters.

5.  Elbow Method Interpretation:

  • The "Optimal Number of Clusters" plot (Elbow curve) suggests 3 clusters, as the curve shows a noticeable bend at k=3k = 3k=3.
  • This validates the choice of k=3k = 3k=3, meaning further increasing clusters may not significantly improve the model.

6.  Conclusion & Business Insight:

  • The dataset has been effectively segmented into three groups.
  • Depending on the application (e.g., customer segmentation, market research), each cluster could represent distinct categories based on common characteristics.
  • Further interpretation requires understanding the feature variables used in clustering.  

    • Related Articles

    • K-Means Clusters Overview

      K-Means is an unsupervised machine learning algorithm used for clustering data into distinct groups based on similarity. It is widely used in pattern recognition, market segmentation, and anomaly detection. How K-Means Works Initialize Clusters: ...

    • K means frequently asked questions

      What is K-Means clustering in? K-Means clustering is an unsupervised machine learning technique available in Sigma Magic that groups similar data points into kkk clusters by minimizing intra-cluster variance. How does Sigma Magic perform K-Means ...

    • Hierarchical Clusters Overview

      Hierarchical clustering is a clustering algorithm that builds a hierarchy of clusters through a tree-like structure called a dendrogram. It is widely used for exploratory data analysis and pattern recognition. Types of Hierarchical Clustering ...

    • Hierarchical Clusters Example

      Problem Statement We have collected data on several different models of cars with respect to several parameters. Create a dendogram that displays the hierarchical relationship between the vehicles. How to perform analysis Step 1: Open Sigma Magic ...

    • Hierarchical clusters frequently asked questions

      What is Hierarchical Clustering? Hierarchical clustering in Sigma Magic is a data analysis technique used to group similar data points into a hierarchy of clusters. It builds a tree-like structure called a dendrogram to visualize relationships ...