10 Scikit-learn Functions Every ML Beginner Should Know

You've cleaned your data. You've visualized it. Now it's time to build models.

Scikit-learn is the most used machine learning library in Python. Here are the 10 functions you'll use in every single ML project — with examples and use cases.

1. train_test_split() — Split Your Data

Always split your data before training. Never train on your test data.

Example:

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

Use case: Keeping 20% of your data aside to test how well your model performs on unseen data.

2. StandardScaler() — Scale Your Features

Most ML algorithms perform better when your data is on the same scale.

Example:

from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

Use case: Scaling age, salary, and experience columns so no single feature dominates the model.

3. LabelEncoder() — Encode Categories

ML models don't understand text. Convert categories to numbers.

Example:

from sklearn.preprocessing import LabelEncoder

le = LabelEncoder()
df["gender_encoded"] = le.fit_transform(df["gender"])
# Male → 1, Female → 0

Use case: Converting gender, city, or department columns into numeric values before training.

4. LogisticRegression() — Classify Yes or No

The simplest classification model. Always start here.

Example:

from sklearn.linear_model import LogisticRegression

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

Use case: Predicting if a customer will churn, if an email is spam, if a patient has a disease.

5. LinearRegression() — Predict Numbers

Use when your target is a number, not a category.

Example:

from sklearn.linear_model import LinearRegression

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

Use case: Predicting house prices, salary based on experience, sales based on ad spend.

6. RandomForestClassifier() — Powerful Classification

More powerful than logistic regression. Works well on most datasets.

Example:

from sklearn.ensemble import RandomForestClassifier

model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train, y_train)
predictions = model.predict(X_test)

Use case: Customer churn prediction, fraud detection, disease classification.

7. accuracy_score() — Measure Your Model

How often is your model correct?

Example:

from sklearn.metrics import accuracy_score

accuracy = accuracy_score(y_test, predictions)
print(f"Accuracy: {accuracy * 100:.2f}%")

Use case: Checking how well your classification model performs on the test set.

8. confusion_matrix() — See Where Your Model Fails

Accuracy alone doesn't tell the full story. This shows you exactly where your model goes wrong.

Example:

from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay

cm = confusion_matrix(y_test, predictions)
ConfusionMatrixDisplay(cm).plot()

Use case: Seeing how many predictions were correct, and what type of errors your model is making.

9. cross_val_score() — Test More Reliably

One train-test split isn't always enough. Cross-validation gives a more honest score.

Example:

from sklearn.model_selection import cross_val_score

scores = cross_val_score(model, X, y, cv=5)
print(f"Average Accuracy: {scores.mean() * 100:.2f}%")

Use case: Getting a reliable accuracy estimate by training and testing on 5 different data splits.

10. classification_report() — Full Model Report

Gives you precision, recall, and F1-score in one shot.

Example:

from sklearn.metrics import classification_report

print(classification_report(y_test, predictions))

Use case: Evaluating your model beyond just accuracy — especially important for imbalanced datasets.

Quick Reference — Save This

One Important Rule

Always scale your test data using the scaler fitted on training data.

# Correct ✅
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

# Wrong ❌
X_test_scaled = scaler.fit_transform(X_test)

Fitting the scaler on test data causes data leakage. Your model will look better than it actually is.

Save this article and come back to it every time you start a new ML project.