Predicting Red Wine Quality

This task involves building a predictive model for red wine quality based on its chemical properties. You are provided with a clean CSV dataset containing various attributes of red wines.

Dataset Details

The target (dependent) variable is:

• quality (integer): A subjective wine quality score on a 1–7 scale.

There are approximately 10 input (independent) variables, all numeric, describing the wine’s chemical properties. Examples include:

• fixed_acidity (float)
• volatile_acidity (float)
• citric_acid (float)
• residual_sugar (float)
• chlorides (float)
• free_sulfur_dioxide (float)
• total_sulfur_dioxide (float)
• density (float)
• pH (float)
• sulphates (float)
• alcohol (float)

Assumptions

• The dataset contains no missing values.
• Each row represents an independent and identically distributed (i.i.d.) wine sample, unless you discover evidence to the contrary during your analysis.

Tasks

1. Exploratory Data Analysis (EDA): Describe your key findings from exploring the dataset, focusing on distributions, outliers, correlations, target imbalance, and non-linearities. List at least three concrete findings and explain how each finding would influence your subsequent modeling choices.
2. Pre-Model Feature Usefulness: Identify which input variables appear most likely to be useful for predicting quality and justify your reasoning. You must mention at least two distinct methods for assessing feature usefulness (e.g., correlation coefficients, mutual information, monotonic trends, or domain-specific insights).
3. Modeling: Develop a model to predict quality. You have the freedom to choose any appropriate machine learning approach. Clearly specify and justify the following:
   • Whether you treat the task as regression, classification, or ordinal classification, and provide a clear rationale for your choice.
   • Your chosen train/validation strategy (e.g., a simple split or cross-validation).
   • The evaluation metric(s) you would use to assess your model's performance.
4. Post-Model Feature Importance: Describe how you would determine which variables are actually most influential in your final model. Propose a method that is appropriate for your chosen model and discuss potential pitfalls associated with that method (e.g., issues with collinearity, data leakage, or bias in impurity-based importances).

Deliverable

Provide a concise write-up detailing your approach and results. Optionally, you may include pseudocode or a code outline, preferably in Python using libraries like pandas and scikit-learn.