File size: 6,387 Bytes
7b54266 4ae907d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 | ---
license: mit
language: en
tags:
- cybersecurity
- malicious-url-detection
- url-classification
- machine-learning
- phishing-detection
pipeline_tag: text-classification
---
# Malicious URL Detection Models
This directory contains trained machine learning models for detecting malicious URLs. The models are trained to classify URLs into four categories:
- **benign**
- **defacement**
- **malware**
- **phishing**
## Model Performance Summary
The following table summarizes the accuracy of each model on the test dataset:
| Model | Accuracy |
|-------|----------|
| **Extra Trees Classifier** | **97%** |
| **Random Forest** | **97%** |
| **Decision Tree** | **96%** |
| **MLP Classifier** | **96%** |
| **XGBoost** | **96%** |
| **Gradient Boosting Classifier** | **94%** |
| **Logistic Regression** | **87%** |
| **SGD Classifier** | **87%** |
| **Adaboost** | **85%** |
| **Gaussian Naive Bayes** | **80%** |
## Detailed Performance Reports
### Adaboost
- **Accuracy:** 0.85
- **Report:**
```
precision recall f1-score support
benign 0.90 0.97 0.93 85778
defacement 0.82 0.76 0.79 19104
malware 0.55 0.74 0.63 6521
phishing 0.68 0.42 0.52 18836
accuracy 0.85 130239
macro avg 0.74 0.72 0.72 130239
weighted avg 0.84 0.85 0.84 130239
```
### Decision Tree
- **Accuracy:** 0.96
- **Report:**
```
precision recall f1-score support
benign 0.97 0.98 0.98 85778
defacement 0.98 0.99 0.98 19104
malware 0.95 0.94 0.95 6521
phishing 0.87 0.85 0.86 18836
accuracy 0.96 130239
macro avg 0.95 0.94 0.94 130239
weighted avg 0.96 0.96 0.96 130239
```
### Extra Trees Classifier
- **Accuracy:** 0.97
- **Report:**
```
precision recall f1-score support
benign 0.97 0.98 0.98 85778
defacement 0.98 0.99 0.99 19104
malware 0.98 0.94 0.96 6521
phishing 0.91 0.86 0.88 18836
accuracy 0.97 130239
macro avg 0.96 0.95 0.95 130239
weighted avg 0.97 0.97 0.97 130239
```
### Gaussian Naive Bayes
- **Accuracy:** 0.80
- **Report:**
```
precision recall f1-score support
benign 0.86 0.90 0.88 85778
defacement 0.67 0.99 0.80 19104
malware 0.63 0.69 0.66 6521
phishing 0.68 0.19 0.29 18836
accuracy 0.80 130239
macro avg 0.71 0.69 0.66 130239
weighted avg 0.80 0.80 0.77 130239
```
### Gradient Boosting Classifier
- **Accuracy:** 0.94
- **Report:**
```
precision recall f1-score support
benign 0.96 0.99 0.97 85778
defacement 0.92 0.97 0.94 19104
malware 0.94 0.80 0.87 6521
phishing 0.89 0.78 0.83 18836
accuracy 0.94 130239
macro avg 0.93 0.88 0.90 130239
weighted avg 0.94 0.94 0.94 130239
```
### Logistic Regression
- **Accuracy:** 0.87
- **Report:**
```
precision recall f1-score support
benign 0.89 0.97 0.93 85778
defacement 0.85 0.95 0.90 19104
malware 0.81 0.69 0.74 6521
phishing 0.77 0.42 0.55 18836
accuracy 0.87 130239
macro avg 0.83 0.76 0.78 130239
weighted avg 0.87 0.87 0.86 130239
```
### MLP Classifier
- **Accuracy:** 0.96
- **Report:**
```
precision recall f1-score support
benign 0.97 0.98 0.98 85778
defacement 0.97 0.97 0.97 19104
malware 0.95 0.90 0.92 6521
phishing 0.88 0.83 0.86 18836
accuracy 0.96 130239
macro avg 0.94 0.92 0.93 130239
weighted avg 0.96 0.96 0.96 130239
```
### Random Forest
- **Accuracy:** 0.97
- **Report:**
```
precision recall f1-score support
benign 0.98 0.98 0.98 85778
defacement 0.98 0.99 0.99 19104
malware 0.98 0.94 0.96 6521
phishing 0.91 0.87 0.89 18836
accuracy 0.97 130239
macro avg 0.96 0.95 0.95 130239
weighted avg 0.97 0.97 0.97 130239
```
### SGD Classifier
- **Accuracy:** 0.87
- **Report:**
```
precision recall f1-score support
benign 0.89 0.96 0.93 85778
defacement 0.83 0.95 0.89 19104
malware 0.79 0.71 0.75 6521
phishing 0.74 0.40 0.52 18836
accuracy 0.87 130239
macro avg 0.81 0.76 0.77 130239
weighted avg 0.86 0.87 0.85 130239
```
### XGBoost
- **Accuracy:** 0.96
- **Report:**
```
precision recall f1-score support
benign 0.97 0.99 0.98 85778
defacement 0.97 0.99 0.98 19104
malware 0.98 0.92 0.95 6521
phishing 0.91 0.84 0.88 18836
accuracy 0.96 130239
macro avg 0.96 0.93 0.95 130239
weighted avg 0.96 0.96 0.96 130239
```
## Usage
To load a model in Python, you can use `joblib` or `pickle`.
### Using joblib
```python
import joblib
# Load the model
model = joblib.load('models/random_forest.pkl')
# Make predictions
prediction = model.predict(X_test)
```
### Using pickle
```python
import pickle
# Load the model
with open('models/random_forest.pkl', 'rb') as f:
model = pickle.load(f)
# Make predictions
prediction = model.predict(X_test)
```
|