Wine Quality Analysis

A comprehensive R script that downloads, cleans, and analyzes the UCI Wine Quality datasets (red & white) and produces a suite of exploratory visualizations, principal component analysis, and a predictive regression model.

---

Table of Contents

1. Project Overview
2. Features
3. Prerequisites
4. Installation
5. Usage
6. Script Breakdown
7. Extending & Customizing
8. Data Sources & Citations
9. License

---

Project Overview

This project provides a single, self-contained R script (Wine_Quality_Analysis.R) that:

• Automatically installs and loads required packages
• Downloads red and white wine quality data from the UCI repository
• Parses semicolon-delimited files with explicit decimal handling
• Merges into one tidy data frame with a type factor
• Creates multiple exploratory plots to uncover chemical and quality patterns
• Performs Principal Component Analysis (PCA) on standardized attributes
• Fits a linear regression predicting wine quality from key predictors
• Outputs summary statistics, diagnostic plots, and fitted vs. actual comparisons

Ideal for data scientists, statisticians, and wine enthusiasts seeking a reproducible pipeline for exploring chemical correlates of perceived wine quality.

---

Features

1. Data Ingestion & Cleaning

   • Downloads CSVs for red & white wines
   • Enforces consistent numeric types via read_delim() with explicit locale

2. Exploratory Visualizations

   • Bar chart of quality counts by type
   • Jittered scatter of alcohol content vs. quality
   • Faceted boxplots for key chemical attributes by quality
   • Correlation heatmap of all numeric variables

3. Dimensionality Reduction (PCA)

   • Scree plot of variance explained
   • Biplot illustrating attribute loadings

4. Predictive Modeling

   • Linear regression of numeric quality on alcohol, sulphates, pH, and type
   • Model summary and residuals vs. fitted diagnostic
   • Scatter of fitted vs. actual quality

---

Prerequisites

• R (version ≥ 4.0)
• Internet connection (to download data)

R Packages

• tidyverse (ggplot2, dplyr, tidyr, readr)
• GGally
• corrplot
• factoextra

All missing packages will be installed automatically by the script.

---

Installation

1. Clone or download this repository:

   git clone https://github.com/yourusername/wine-quality-analysis.git
   cd wine-quality-analysis

2. Ensure R (≥4.0) is on your PATH.

No additional build steps required.

---

Usage

Open an R console or RStudio in the project directory and run:

source("Wine_Quality_Analysis.R")

The script will:

• Print basic counts of red vs. white wines
• Render each plot in sequence
• Display PCA diagnostics
• Print regression summary
• Show diagnostic plots and final fitted vs. actual chart

---

Script Breakdown

1. Setup

   • Defines package list, installs missing ones
   • Loads libraries

2. Data Loading

   • Uses read_delim() with locale(decimal_mark=".")
   • Applies a col_types spec to ensure all measurement columns are double()

3. Data Wrangling

   • Adds a type factor (red/white)
   • Converts quality to an ordered factor

4. Exploratory Plots

   • p1: Bar chart of quality counts
   • p2: Jitter scatter of alcohol vs. quality
   • p3: Faceted boxplots for pH, residual sugar, citric acid, sulphates

5. Correlation Analysis

   • Correlation matrix heatmap of all numeric variables

6. PCA

   • Scree plot of variance explained
   • Biplot of the first two principal components

7. Regression

   • Linear model: quality ~ alcohol + sulphates + pH + type
   • Summary printed to console
   • Residuals vs. fitted plot
   • Fitted vs. actual scatter

---

Extending & Customizing

• Add predictors (e.g., residual sugar, free sulfur dioxide) to the regression
• Try nonlinear models (e.g., random forest, gradient boosting)
• Tune PCA by selecting different scaling or component count
• Export plots by inserting ggsave() calls after each print()

---

Data Sources & Citations

• Wine Quality Data UCI Machine Learning Repository:

  • Red wine: https://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-red.csv
  • White wine: https://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-white.csv

• R & Packages

  • R Core Team (2023). R: A language and environment for statistical computing.
  • Hadley Wickham et al., for tidyverse, GGally, corrplot, factoextra.

---

License

Distributed under the MIT License. See LICENSE for details.