README.md

---
license: mit
library_name: sklearn
tags:
- tabular-classification
- sklearn
- phishing
- onnx
model_format: pickle
model_file: model.pkl
widget:
- structuredData:
    domain_age:
    - 11039.0
    - -1.0
    - 5636.0
    domain_registration_length:
    - 3571.0
    - 0.0
    - 208.0
    google_index:
    - 0.0
    - 0.0
    - 0.0
    nb_hyperlinks:
    - 97.0
    - 168.0
    - 52.0
    page_rank:
    - 5.0
    - 2.0
    - 10.0
    ratio_extHyperlinks:
    - 0.030927835
    - 0.220238095
    - 0.442307692
    ratio_extRedirection:
    - 0.0
    - 0.378378378
    - 0.0
    ratio_intHyperlinks:
    - 0.969072165
    - 0.779761905
    - 0.557692308
    safe_anchor:
    - 25.0
    - 24.32432432
    - 0.0
    status:
    - legitimate
    - legitimate
    - legitimate
    web_traffic:
    - 178542.0
    - 0.0
    - 2.0
inference: false
pipeline_tag: tabular-classification
---

# Model Description


The model predicts the probability that a URL is a phishing site using a list of features.

- **Model type:** Traditional machine learning
- **Task:** Tabular classification (Binary)
- **License:**: MIT
- **Repository:** https://github.com/pirocheto/phishing-url-detection


## Evaluation

| Metric    |    Value |
|-----------|----------|
| accuracy  | 0.945652 |
| f1-score  | 0.945114 |
| precision | 0.951996 |
| recall    | 0.938331 |

# How to Get Started with the Model


Using pickle in Python is discouraged due to security risks during data deserialization, potentially allowing code injection.
It lacks portability across Python versions and interoperability with other languages.

Instead, we recommend using the ONNX model, which is more secure. 
It is half the size and almost twice as fast compared to the pickle model. 
Additionally, it can be utilized by languages supported by the [ONNX runtime](https://onnxruntime.ai/docs/get-started/) (see below for an example using JavaScript).


## With ONNX (recommanded)

### Python

```python
import onnxruntime
import pandas as pd
from huggingface_hub import hf_hub_download

REPO_ID = "pirocheto/phishing-url-detection"
FILENAME = "model.onnx"

# Initializing the ONNX Runtime session with the pre-trained model
sess = onnxruntime.InferenceSession(
    hf_hub_download(repo_id=REPO_ID, filename=FILENAME),
    providers=["CPUExecutionProvider"],
)

# Defining a list of URLs with characteristics
data = [
    {
        "url": "https://www.rga.com/about/workplace",
        "nb_hyperlinks": 97,
        "ratio_intHyperlinks": 0.969072165,
        "ratio_extHyperlinks": 0.030927835,
        "ratio_extRedirection": 0,
        "safe_anchor": 25,
        "domain_registration_length": 3571,
        "domain_age": 11039,
        "web_traffic": 178542,
        "google_index": 0,
        "page_rank": 5,
    },
]

# Converting data to a float32 NumPy array
df = pd.DataFrame(data).set_index("url")
inputs = df.to_numpy(dtype="float32")

# Using the ONNX model to make predictions on the input data
probas = sess.run(None, {"X": inputs})[1]

# Displaying the results
for url, proba in zip(data, probas):
    print(f"URL: {url['url']}")
    print(f"Likelihood of being a phishing site: {proba[1] * 100:.2f}%")
    print("----")

# Expected output:
# URL: https://www.rga.com/about/workplace
# Likelihood of being a phishing site: 0.89%
# ----

```

### JavaScript

```javascript
const ort = require('onnxruntime-node');

const data = [
    {
        "url": "http://rapidpaws.com/wp-content/we_transfer/index2.php?email=/",
        "nb_hyperlinks": 1,
        "ratio_intHyperlinks": 1,
        "ratio_extHyperlinks": 0,
        "ratio_extRedirection": 0,
        "safe_anchor": 0,
        "domain_registration_length": 338,
        "domain_age": 0,
        "web_traffic":1853,
        "google_index": 1,
        "page_rank": 2,
    },
];

async function main() {
    try {
        // Make sure you have downloaded the model.onnx
        // Creating an ONNX inference session with the specified model
        const model_path = "./models/model.onnx";
        const session = await ort.InferenceSession.create(model_path);
        
        // Creating an ONNX tensor from the input data
        const inputs = data.map(url => Object.values(url).slice(1));
        const flattenInputs = inputs.flat();
        const tensor = new ort.Tensor('float32', flattenInputs, [inputs.length, 10]);
        
        // Executing the inference session with the input tensor
        const results = await session.run({"X": tensor});
        const probas = results['probabilities'].data;
        
        // Displaying results for each URL
        data.forEach((url, index) => {
            const proba = probas[index * 2 + 1];
            const percent = (proba * 100).toFixed(2);
            
            console.log(`URL: ${url.url}`);
            console.log(`Likelihood of being a phishing site: ${percent}%`);
            console.log("----");
        });
        

    } catch (e) {
        console.log(`failed to inference ONNX model: ${e}.`);
    }
};

main();

// Expected output:
// URL: https://www.rga.com/about/workplace
// Likelihood of being a phishing site: 0.89%
// ----

```