What is Machine Learning?

# Traditional programming: rules are written by humans
def classify_email(email):
    if "buy now" in email.lower() or "click here" in email.lower():
        return "spam"
    if "meeting" in email.lower() or "agenda" in email.lower():
        return "important"
    return "normal"
# Problem: brittle, needs constant manual updates for new spam tactics

# Machine Learning: rules are LEARNED from data
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.pipeline import Pipeline

# Training data: examples the model learns from
emails = [
    "Buy now! Click here for amazing deals!",
    "Hi, can we schedule a meeting for Monday?",
    "FREE PRIZE you have been selected click immediately",
    "Please review the attached agenda for tomorrow",
]
labels = ["spam", "ham", "spam", "ham"]

# Build and train the model
model = Pipeline([
    ("vectorizer", TfidfVectorizer()),
    ("classifier", MultinomialNB()),
])
model.fit(emails, labels)

# Predict on new, unseen data
new_email = "Limited time offer! Get 90% off today!"
prediction = model.predict([new_email])
print(f"Prediction: {prediction[0]}")  # spam

# Key advantage: automatically adapts to new spam patterns with more training data