Key Performance Indicators
Total Predictions
284,807
Transactions analyzed
Fraud Detection Rate
0.17%
Transactions flagged as fraud
Model Accuracy
1.0%
Overall model performance
+2.1%vs previous version
Avg Confidence
98.2%
Average prediction confidence
Transaction Analytics
Transaction Trends
Daily transaction volume and fraud detection rates
Risk Distribution
Breakdown of transactions by risk level
High36 (23.1%)
Medium34 (21.8%)
Low86 (55.1%)
Total
156
High Risk
36
Low Risk
86
Model Performance
Confusion Matrix
Model performance breakdown (218 predictions)
Predicted: Fraud
Predicted: Normal
Actual: Fraud
True Positive
44
20.2%
Correctly identified
fraud transactions
False Negative
2
0.9%
Missed fraud
(false negative)
Actual: Normal
False Positive
23
10.6%
False alarm
(false positive)
True Negative
149
68.3%
Correctly identified
normal transactions
Performance Metrics
Accuracy
88.5%
Overall correctness
Precision
65.7%
Fraud prediction accuracy
Recall
95.7%
Fraud detection rate
F1-Score
77.9%
Harmonic mean
Accuracy: 88.5% of all predictions were correct
Precision: 65.7% of fraud predictions were actual fraud
Recall: 95.7% of actual fraud cases were detected
F1-Score: 77.9% balanced performance metric
Feature Importance
Top 15 most important features for fraud detection
Temporal Features
Amount Features
PCA Features (V1-V28)
Total Features: 30
Showing Top: 15
Cumulative Importance: 74.2%
Recent Predictions
Latest Transactions
Loading recent predictions...
Loading...
Model Information
Model Details
Type:lightgbm
Version:1.0.0
Format:onnx
Features:30
Performance Metrics
Accuracy:98.2%
Precision:7.8%
Recall:89.8%
F1-Score:14.3%
Dataset Statistics
Total:284,807
Fraud:492
Normal:284,315
Fraud Rate:0.17%