Embedded Files
Abstract
Abstract
In this project, I utilize clustering, an unsupervised machine learning technique, to categorize wines based on their physicochemical properties. The goal of clustering is to group data points with similar traits together and keep dissimilar ones apart (Seif, 2018). By applying this method to wines, I aim to identify groups that share similar chemical attributes, enabling personalized wine recommendations based on their chemical profiles.
Background
Background
Wine flavors are influenced by grape varieties, the regions where they're grown, and their vintage year. While wines are traditionally classified based on these factors, the subtler nuances in flavor are often a result of variations in physicochemical properties such as pH, density, residual sugars, and sulphates (Cortez et al., 2009; Notman, 2018). This understanding suggests that wines with similar chemical make-ups might share taste profiles. Therefore, wines from the same region can have different flavors, and wines from different regions might taste similar.
In my research, I focus on analyzing the chemical similarities among wines to enhance my understanding of their flavor profiles. If a particular wine with specific chemical properties appeals to someone, they might also enjoy other wines from the same cluster with similar chemical compositions. This research can inform personal wine choices and deepen my understanding of wine characteristics.
Objective
Objective
My objective is to conduct a clustering analysis on red wines based on their physicochemical properties. This study aims to uncover associations within these properties, enhancing my knowledge and approach to wine selection and appreciation.
Data
Data
I use the red wine quality dataset from the UCI Machine Learning Repository (http://www3.dsi.uminho.pt/pcortez/wine/winequality.zip) for this analysis. Although it is traditionally used for classification tasks, the dataset's 11 physicochemical features and 4898 observations make it suitable for clustering analysis. This project allows me to explore the relationships between the various physicochemical attributes of red wines.
Colab Notebook
Colab Notebook
PDF
PDF
Wine Clustering Based on Chemical Composition.pdfPage updated
Report abuse