wine-quality-ml-model

The ONE rule that decides everything 🧠

If the target is a number and the distance between values matters → REGRESSION If the target is a category and distance does NOT matter → CLASSIFICATION

Aspect	Regression	Classification
Target	numeric score	category
Output	1 neuron	N neurons
Loss	MSE / MAE	CrossEntropy
Prediction	5.6	"medium"
Error meaning	distance matters	distance ignored
Natural fit here	✅ YES	⚠️ Optional

📌 MSELoss expects FLOAT targets, not integers (Long).

Unsupervised learning & data exploration

This answers:

“What if I don’t have labels?”

Learn

Clustering (KMeans, DBSCAN)

PCA / dimensionality reduction

Anomaly detection

With your wine data

Cluster wines by chemistry

Visualize red vs white separation

Detect outlier wines

📌 This is common in real datasets, where labels are missing or unreliable.