Wine Quality Profiling using K-Means Clustering (Unsupervised Learning)-by Cherinet Bekele

[chere-collab]

Wine Quality Profiling and Classification using K-Means Clustering

1. Data Cleaning & Preparation
   Encoding: The model converted the type column from text into numerical values (0 for Red, 1 for White).
   Scaling: The model standardized all numerical columns (such as alcohol, density, and residual sugar) to ensure they all contribute equally to the distance calculation.

2. Exploratory Visualization
   Correlation Analysis: The model analyzed the relationships between all chemical columns (like pH and fixed acidity) using a heatmap.
   Initial Plotting: The model visualized the distribution of specific columns, such as alcohol and density, to see how the wine types naturally separated.

3.Determining the Optimal Clusters (K)Mathematical Calculation: The model computed the WCSS (error) for the entire set of features across different cluster counts.

4. Executing K-Means Clustering
   Partitioning: The model sorted all wines into three distinct groups based on the multi-dimensional values of all 12 chemical columns.
   Centroid Calculation: The model calculated the arithmetic mean for every column (e.g., average volatile acidity, sulphates, and quality) to establish the new center of each cluster.

5.New Data Prediction
Classification: The model processed a new set of wine features through the same scaling rules used for the original columns.
Assignment: The model predicted the cluster by comparing the new wine's chemical measurements against the final means (centroids) of each established group.