Alidzhon Aminov

Classification with NN and Tree-based Machine Learning Models: Wine Quality Dataset

In this project, I did classification of wine quality using several Tree-based Machine Learning models and simple Neural Network model:

• Gradient Boosting Machines (GBM)
• XGBoost
• RandomForest
• Neural Network

The key steps included:

• Data Preparation and Preprocessing:
• Data Exploration: To check the types of data within the dataset, identify missing values, and detect duplicates, to ensure the dataset is clean for analysis..
• Feature Selection After an exploratory data analysis, I selected features based on their correlation with wine quality and domain knowledge.
• Data Transformation Applied log transformation to certain skewed features to normalize their distribution, improving model performance.
• Model Training and Evaluation: Tuned with GridSearchCV to find the best parameters.
• Results and Analysis: On the one hand, RandomForest performed the best compared to other tree-based machine learning methods (XGBoost, Gradient Boosting Machines (GBM)), demonstrating its robustness and effectiveness in handling variability. On the other hand, the Neural Network model with a simple architecture showed superior performance.
  However, the choice of which model to use depends on the importance of the model's interpretability, deployment requirements, and operational considerations.