<< back to Guides

AI Core Concepts (Part 11): Supervised Learning

Supervised Learning is a type of machine learning where models are trained on labeled data—each input is associated with a known output (label). The model learns to map inputs to outputs by minimizing a loss function.

1. Core Idea

Learn a function f(x) ≈ y from examples of (x, y) pairs.
The model predicts outputs on new unseen inputs.
Loss is computed based on how close predictions are to true labels.

2. Types of Supervised Learning

🔹 Classification

Goal: Predict a discrete label.
Examples: Spam detection, image classification, sentiment analysis

from sklearn.linear_model import LogisticRegression

model = LogisticRegression()
model.fit(X_train, y_train)
predictions = model.predict(X_test)

🔹 Regression

Goal: Predict a continuous value.
Examples: Price prediction, temperature forecasting, stock values

from sklearn.linear_model import LinearRegression

regressor = LinearRegression()
regressor.fit(X_train, y_train)
y_pred = regressor.predict(X_test)

3. Common Algorithms

Task	Algorithms
Classification	Logistic Regression, SVM, Random Forest, k-NN, Neural Networks
Regression	Linear Regression, SVR, XGBoost, Decision Trees, Ridge/Lasso
Universal	Gradient Boosting, Neural Nets, Transformers (with labels)

4. Supervised Learning Pipeline

Data Preparation – clean, normalize, encode
Train/Test Split – evaluate generalization
Model Selection – pick based on data & task
Training – learn parameters on training data
Evaluation – use metrics (accuracy, MSE, F1)
Tuning – hyperparameter optimization
Deployment – inference on new data

5. Example: End-to-End Classification

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score

# Load and split data
data = load_iris()
X_train, X_test, y_train, y_test = train_test_split(data.data, data.target, test_size=0.2)

# Train
clf = RandomForestClassifier()
clf.fit(X_train, y_train)

# Predict and evaluate
y_pred = clf.predict(X_test)
print("Accuracy:", accuracy_score(y_test, y_pred))

6. Key Evaluation Metrics

Task	Metrics
Classification	Accuracy, Precision, Recall, F1
Regression	Mean Squared Error (MSE), R² score
All Tasks	Cross-validation, confusion matrix

7. When to Use Supervised Learning

✅ Use when:

You have labeled data
The problem is well-defined
You want to predict something specific

❌ Avoid when:

You only have unlabeled data (consider unsupervised learning)
Task involves sequential decisions (consider reinforcement learning)

📚 Further Resources

<< back to Guides