How danny777-glitch/phishing-email-detection works
PhishGuard AI is a Streamlit phishing-email detection app backed by a scikit-learn model. Users can scan one email or upload datasets for batch predictions and CSV export; a separate training script builds and serializes the model. The app excerpt shows model loading, feature preparation, and evaluation-matrix code, while batch processing and some scan-result details are documented but partly unsampled.
Main entry points: app.py Streamlit workspace · phishing_detector.pkl Serialized detector · train.py Training script · phishing_dataset.xlsx Email dataset
Components 11
Detection App
- Streamlit workspace
app.py - Email scanner
app.py - Batch processor
app.py - Confusion matrix
app.py
Model Lifecycle
- Training script
train.py - Feature pipeline
train.py - Naive Bayes model
train.py - Serialized detector
phishing_detector.pkl
Inputs and Reports
- Email dataset
phishing_dataset.xlsx - CSV audit report
Other components
- Email analyst
Connections 13
- Email analyst → Streamlit workspace: uses
README.md - Streamlit workspace → Email scanner: opens scanner
README.md - Streamlit workspace → Batch processor: opens batch
README.md - Streamlit workspace → Confusion matrix: shows evaluation
app.py - Training script → Email dataset: reads
train.py - Training script → Feature pipeline: configures
train.py - Feature pipeline → Naive Bayes model: feeds features
train.py - Training script → Serialized detector: serializes
train.py - Streamlit workspace → Serialized detector: loads
app.py - Email scanner → Serialized detector: predicts with
README.md - Batch processor → Serialized detector: predicts with
README.md - Batch processor → CSV audit report: exports
README.md - Confusion matrix → Email dataset: evaluates labels
app.py