Credit Card Fraud Detection

Goal

Fraud datasets are usually highly imbalanced.

You will:

• Explore class imbalance
• Check feature distributions
• Build a baseline model evaluation plan

Analysis pipeline

diagram Fraud detection analysis pipeline mermaid

  flowchart LR
  A["Raw transactions"] --> B["Check imbalance
(value_counts)"]
  B --> C["Compare
(feature dist. by class)"]
  C --> D["Visualize
(kde/overlap)"]
  D --> E["Pick metric
(precision/recall/AUC)"]

From raw transactions to a metric that actually matters for imbalanced data.

From the book

McKinney’s frequency-table pattern — value_counts() or Counter.most_common() — is exactly how you should first look at a fraud label. df["fraud"].value_counts(normalize=True) often reveals something like 99.8% legit vs 0.2% fraud. A model that just predicts “never fraud” would score 99.8% accuracy while being completely useless — which is why accuracy alone is the wrong metric here.

Step 1: Load

Load

import pandas as pd
 
df = pd.read_csv("data/fraud.csv")
print(df.shape)
print(df.head())

Step 2: Class imbalance

Imbalance

print(df["fraud"].value_counts())
print(df["fraud"].value_counts(normalize=True))

Step 3: Visualize distributions

Feature distribution

import seaborn as sns
import matplotlib.pyplot as plt
 
# Example: compare one feature by class
feature = "amount"
 
plt.figure(figsize=(7, 4))
sns.kdeplot(data=df, x=feature, hue="fraud", common_norm=False)
plt.title(f"{feature} distribution by class")
plt.tight_layout()
plt.show()

Visualize it

sketch Class imbalance: legit vs fraud p5.js Pause Restart

The fraud bar is intentionally tiny - that is the whole challenge of this dataset.

Step 4: Baseline modeling note

For fraud, accuracy is misleading.

Prefer:

• Precision / Recall
• F1 score
• ROC-AUC
• PR-AUC

Deliverable

• How imbalanced is the dataset?
• Which features differ between classes?
• What metric will you optimize?

🧪 Try It Yourself

Exercise 1 – Measure class imbalance

pch.exerciseTag pch.exerciseTitle

Exercise 2 – Compare a feature by class

pch.exerciseTag pch.exerciseTitle

Exercise 3 – Why accuracy misleads

pch.exerciseTag pch.exerciseTitle