Project 1: Spam Detection with scikit-learn

This is one of the most important beginner ML projects. It teaches the full supervised-learning loop.

What You Will Build

• binary text classifier (spam vs ham)
• reusable preprocessing + model pipeline
• evaluation report with precision/recall/F1

Core Implementation

from sklearn.model_selection import train_test_split
from sklearn.pipeline import Pipeline
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import classification_report

X_train, X_test, y_train, y_test = train_test_split(
    messages, labels, test_size=0.2, random_state=42, stratify=labels
)

clf = Pipeline([
    ("tfidf", TfidfVectorizer(ngram_range=(1, 2), min_df=2)),
    ("model", LogisticRegression(max_iter=2000))
])

clf.fit(X_train, y_train)
preds = clf.predict(X_test)
print(classification_report(y_test, preds))

Real-World Improvements

• add class balancing when spam ratio is low
• tune thresholds for business risk (false positives vs false negatives)
• inspect misclassified samples and update preprocessing rules

Deliverables

1. Notebook with baseline + improved model
2. Confusion matrix and metric explanation
3. Small README with deployment considerations