This repository contains two main projects:
- GradeClassifier: A project focused on predicting the quality of wines using various physicochemical properties and applying different preprocessing techniques, exploratory data analysis (EDA), and machine learning models.
- ColorClassifier: This project predicts the color of the wine (red or white) based on the wine's properties.
The GradeClassifier project is dedicated to predicting the quality of wines based on various features such as acidity, alcohol content, pH, etc. The workflow includes the following steps:
-
Exploratory Data Analysis (EDA): A thorough analysis to understand feature distributions, correlations between variables, and their relation to the target variable (quality).
-
Preprocessing:
- Normalization: Scaling all features using MinMaxScaler.
- Box-Cox Transformation: Applied to ensure a more normal distribution for certain features.
- Feature Encoding: Transforming categorical features where necessary.
-
Models Used:
- Logistic Regression (Multinomial)
- ElasticNet Regression (Penalized Regression)
- Support Vector Machine (SVM)
- Random Forest
- Gradient Boosting (Optional depending on performance comparison)
- XGBoost (for performance tuning)
- Neural Network (2 hidden layers)
-
Evaluation Metrics:
- Accuracy
- Precision
- Recall
- F1-Score
- Confusion Matrix: Used to visualize and evaluate model performance.
The ColorClassifier project predicts whether a wine is red or white based on its physicochemical properties. This project is simpler and focuses on color classification using basic machine learning techniques.
The GradeClassifier project is the main focus of this repository, providing a detailed analysis and comparison of different machine learning models for wine quality prediction. The ColorClassifier, though simpler, showcases basic classification methods for predicting wine color based on its properties.