DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Review of classification
methods for fraud
detection
Charlotte Werger
Data Scientist
DataCamp Fraud Detection in Python
What is classification?
Goal of classification: Use known fraud cases to train a model to
recognise new fraud cases
Examples:
Email Spam/Not spam
Transaction online fraudulent Yes/No
Tumor Malignant/Benign?
Variable to predict: y ∈ 0, 1
0: Negative class ("majority" normal cases)
1: Positive class ("minority" fraud cases)
DataCamp Fraud Detection in Python
Classification methods commonly used for fraud detection
Logistic Regression
DataCamp Fraud Detection in Python
Classification methods commonly used for fraud detection
Neural Network
DataCamp Fraud Detection in Python
Classification methods commonly used for fraud detection
Decision trees
Random Forests
DataCamp Fraud Detection in Python
Decision Trees and Random Forests
Random forests are a collection of trees on random subsets of
features
DataCamp Fraud Detection in Python
Random Forests for fraud detection
from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier(random_state=42)
model.fit(X_train, y_train)
predicted = model.predict(X_test)
print (metrics.accuracy_score(y_test, predicted))
0.991324200913242
DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Let's practice!
DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Measuring fraud
detection performance
Charlotte Werger
Data Scientist
DataCamp Fraud Detection in Python
Accuracy isn't everything
Throw accuracy out of the window when working on fraud detection
problems
DataCamp Fraud Detection in Python
False positives, false negatives and actual fraud caught
DataCamp Fraud Detection in Python
Precision Recall trade-off
DataCamp Fraud Detection in Python
Obtaining performance metrics
# Import the packages
from sklearn.metrics import precision_recall_curve
from sklearn.metrics import average_precision_score
# Calculate average precision and the PR curve
average_precision = average_precision_score(y_test, predicted)
# Obtain precision and recall
precision, recall, _ = precision_recall_curve(y_test, predicted)
DataCamp Fraud Detection in Python
Precision-Recall Curve
DataCamp Fraud Detection in Python
ROC curve to compare algorithms
DataCamp Fraud Detection in Python
Confusion matrix and classification report
from sklearn.metrics import classification_report, confusion_matrix
# Obtain predictions
predicted = model.predict(X_test)
# Print classification report using predictions
print(classification_report(y_test, predicted))
precision recall f1-score support
0.0 0.99 1.00 1.00 2099
1.0 0.96 0.80 0.87 91
avg / total 0.99 0.99 0.99 2190
# Print confusion matrix using predictions
print(confusion_matrix(y_test, predicted))
[[2096 3]
[ 18 73]]
DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Let's practice!
DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Adjusting your
algorithms for fraud
detection
Charlotte Werger
Data Scientist
DataCamp Fraud Detection in Python
Balanced weights
model = RandomForestClassifier(class_weight='balanced')
model = RandomForestClassifier(class_weight='balanced_subsample')
model = LogisticRegression(class_weight='balanced')
model = SVC(kernel='linear', class_weight='balanced', probability=True)
DataCamp Fraud Detection in Python
Hyperparameter tuning for fraud detection
model = RandomForestClassifier(class_weight={0:1,1:4},random_state=1)
model = LogisticRegression(class_weight={0:1,1:4}, random_state=1)
model = RandomForestClassifier(n_estimators=10,
criterion=’gini’,
max_depth=None,
min_samples_split=2,
min_samples_leaf=1,
max_features=’auto’,
n_jobs=-1, class_weight=None)
DataCamp Fraud Detection in Python
Using GridSearchCV
from sklearn.model_selection import GridSearchCV
# Create the parameter grid
param_grid = {
'max_depth': [80, 90, 100, 110],
'max_features': [2, 3],
'min_samples_leaf': [3, 4, 5],
'min_samples_split': [8, 10, 12],
'n_estimators': [100, 200, 300, 1000]
}
# Define which model to use
model = RandomForestRegressor()
# Instantiate the grid search model
grid_search_model = GridSearchCV(estimator = model,
param_grid = param_grid, cv = 5,
n_jobs = -1, scoring='f1')
DataCamp Fraud Detection in Python
Finding the best model with GridSearchCV
# Fit the grid search to the data
grid_search_model.fit(X_train, y_train)
# Get the optimal parameters
grid_search_model.best_params_
{'bootstrap': True,
'max_depth': 80,
'max_features': 3,
'min_samples_leaf': 5,
'min_samples_split': 12,
'n_estimators': 100}
# Get the best_estimator results
grid_search.best_estimator_
grid_search.best_score_
DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Let's practice!
DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Using ensemble
methods to improve
fraud detection
Charlotte Werger
Data Scientist
DataCamp Fraud Detection in Python
What are Ensemble Methods: Bagging versus Stacking
DataCamp Fraud Detection in Python
Stacking Ensemble Methods
DataCamp Fraud Detection in Python
Why use ensemble methods for fraud detection
Ensemble methods:
Are robust
Can help you avoid overfitting
Can typically improve prediction performance
Are a winning formula at prestigious Kaggle competitions
DataCamp Fraud Detection in Python
Voting Classifier
from sklearn.ensemble import VotingClassifier
clf1 = LogisticRegression(random_state=1)
clf2 = RandomForestClassifier(random_state=1)
clf3 = GaussianNB()
ensemble_model = VotingClassifier(estimators=[('lr', clf1),
('rf', clf2), ('gnb', clf3)], voting='hard')
ensemble_model.fit(X_train, y_train)
ensemble_model.predict(X_test)
VotingClassifier(estimators=[('lr', clf1), ('rf', clf2),
('gnb', clf3)], voting='soft', weights=[2,1,1])
DataCamp Fraud Detection in Python
Reliable labels for fraud detection
DataCamp Fraud Detection in Python
FRAUD DETECTION IN PYTHON
Let's practice
