K-Means Clustering Optimization for Cardiotocography Dataset

Task Description

The goal of this task is to determine the best grouping for the Cardiotocography dataset using the k-means clustering algorithm. The task requirements include:

• Analyzing the dataset, excluding the "Class" and "NSP" fields.
• Finding the best grouping with a maximum of 15 clusters, considering the reference grouping with 10 groups based on the "Class" attribute.
• Performing a minimum of 10 tests to achieve better results than the default k-means algorithm with 10 groups and without data preprocessing.

Dataset Description

The Cardiotocography dataset consists of 2126 fetal cardiotocograms (CTGs) with various measurements of fetal heart rate (FHR) and uterine contraction (UC) features. It is a multivariate dataset with 23 real-valued attributes.

Script Explanation

The provided script is written in R and performs the following steps:

1. Data Loading and Preparation: Downloads and reads the Cardiotocography dataset, excluding the unnecessary fields.

2. K-Means Clustering: Applies the k-means algorithm with default parameters and calculates the accuracy.

3. Data Scaling: Performs data scaling using the scale() function and applies k-means clustering to the scaled data.

4. Finding Optimal Number of Clusters: Tests different numbers of clusters and plots the within-cluster sum of squares (WSS) to determine the optimal number.

5. Experimentation: Conducts a series of tests by varying the number of clusters and reports the accuracy results.

6. Conclusion: Summarizes the findings and suggests considering alternative clustering algorithms for further improvement.

Getting Started

To run the script and perform k-means clustering optimization on the Cardiotocography dataset, follow these steps:

1. Download the dataset file from the provided URL.
2. Ensure that R and the required libraries are installed.
3. Run the script in an R environment or script editor.
4. Analyze the results and refer to the generated outputs for clustering accuracy and optimization attempts.

Contributions and Future Enhancements

Contributions and enhancements to this project could include:

• Refactoring the code for better readability and maintainability.
• Adding alternative clustering algorithms for comparison.
• Implementing additional evaluation metrics.
• Creating visualizations to better understand the dataset and clustering results.
• Providing alternative optimization techniques or preprocessing methods.

Collaboration and further development of this project will contribute to the understanding and application of clustering algorithms in data analysis.

Happy clustering!