WineDataProject

Description: In this small project I took a look at the field of Data Science.

Technologies Used:

• Written in Python

Libraries Used:

• pandas
• matplotlib
• numpy
• seaborn
• scikit-learn (sklearn)

Quality Distribution

• File: Quality Distribution
• Findings:
  • Created graphs to represent the distribution of wine quality for both red and white wine.
  • Observed clear differences between red and white wine qualities, with white wine generally having higher quality ratings.

Alcohol Quality

• File: Alcohol Quality
• Findings:
  • Plotted the relationship between quality and alcohol content.
  • Discovered a clear connection between higher alcohol levels and higher quality ratings for both red and white wines.

Sweet Quality

• File: Sweet Quality
• Findings:
  • Analyzed the relationship between sweet and dry wines and their quality distribution.
  • Found that white wines are generally sweeter than red wines.
  • Identified no strong positive or negative relationship between sweetness and quality, although there's a weak positive correlation for dry white wines.

Correlation Matrix

• File: Correlation Matrix
• Findings:
  • Utilized heatmap visualization to understand the correlation between different variables and wine quality.
  • Discovered several strong correlations, such as between density and residual sugar in white wines, and between alcohol content and density in both red and white wines.

Machine Learning

• File: Machine Learning
• Findings:
  • Applied logistic regression for binary classification to predict wine quality.
  • Achieved fairly positive results, with mean cross-validation scores suggesting accurate predictions of wine quality.
  • Tested model with varying thresholds and found that lower thresholds resulted in greater accuracy, particularly with a threshold of 4 achieving near-perfect prediction accuracy.
  • Highlighted the importance of variable selection and testing in achieving accurate predictions.