Clustering performance comparison using K-means and expectation maximization algorithms.

Area of Science:

• Data Mining
• Machine Learning
• Statistical Analysis

Background:

• Clustering is a key unsupervised data mining technique for grouping similar data points.
• K-means and Expectation Maximization (EM) are prominent clustering algorithms.
• Logistic regression is a statistical method for predicting event occurrence but may lack accuracy in data classification.

Purpose of the Study:

• To apply logistic regression analysis to EM and K-means clustering methods.
• To assess the quality of red wine using these combined techniques.
• To propose a method for improving the accuracy of classification results.

Main Methods:

• Utilized K-means clustering for data segmentation.
• Employed Expectation Maximization (EM) algorithm for cluster analysis.
• Integrated logistic regression with clustering outputs for quality assessment.

Main Results:

• Demonstrated the application of logistic regression on EM clusters and K-means clusters.
• Evaluated the effectiveness of the proposed method in red wine quality assessment.
• Achieved enhanced accuracy in classification results through the integrated approach.

Conclusions:

• The combination of logistic regression with clustering techniques like EM and K-means can improve data classification accuracy.
• This approach offers a robust method for quality assessment tasks, exemplified by red wine analysis.
• The proposed method provides a reliable strategy for ensuring accurate classification in data mining applications.