Wine Type Classification

🧪 Wine Type Classification from the Wine Dataset

Problem Description

The goal is to build a classification model that determines the type of wine based on its chemical characteristics.

Each sample is characterized by 13 numerical attributes that describe chemical and physical properties of the wine, and the label (target) indicates the wine variety from which it originates.

This type of problem belongs to the multi-class classification category.

🔹 Features

• alcohol
• malic_acid
• ash
• alcalinity_of_ash
• magnesium
• total_phenols
• flavanoids
• nonflavanoid_phenols
• proanthocyanins
• color_intensity
• hue
• od280/od315_of_diluted_wines
• proline

The dataset comes from the original UCI Machine Learning Repository collection:
https://archive.ics.uci.edu/ml/datasets/Wine

📘 Input File Structure

train.csv

Contains all 13 features columns plus the column:

• target – represents the wine class (variety).
  Possible values are 1, 2 and 3.

Example:

   SampleID  alcohol  malic_acid  ...  od280/od315_of_diluted_wines  proline  target
0        37    13.28        1.64  ...                          2.78    880.0       0
1        31    13.73        1.50  ...                          2.71   1285.0       0
2        27    13.39        1.77  ...                          3.22   1195.0       0
3        13    13.75        1.73  ...                          2.90   1320.0       0
4       149    13.32        3.24  ...                          1.62    650.0       2

test.csv

Contains the same columns without target, but includes SampleID.

Example:

   SampleID  alcohol  malic_acid  ...   hue  od280/od315_of_diluted_wines  proline
0        11    14.10        2.16  ...  1.25                          3.17   1510.0
1       135    12.51        1.24  ...  0.75                          1.51    650.0
2        29    13.87        1.90  ...  1.25                          3.40    915.0
3       122    11.56        2.05  ...  0.93                          3.69    465.0
4        63    13.67        1.25  ...  1.23                          2.46    630.0

📤 Submission

The output file (submission.csv) must contain exactly two columns:

• SampleID
• label – the label predicted by the model (1, 2 or 3)

Example:

⚙️ Evaluation

Model evaluation will be performed using the following metric:

• Macro F1-score

This metric is suitable for multi-class classification because it gives equal weight to each class, regardless of the number of examples in each.

General formula:

where C is the number of classes, and F1_i is the F1 score for class i.

The final score is expressed as a percentage (0–100), rounded to two decimal places.

📊 Source

The dataset comes from the original collection:
UCI Machine Learning Repository – Wine Data Set