Phishing URL Detection System

Detects malicious URLs in real time using machine learning and a REST API.

How it works

1. User sends a URL to the Flask /predict endpoint
2. 11 features are extracted (length, dots, hyphens, IP check, keywords...)
3. Random Forest model classifies it as Legitimate or Phishing
4. API returns the result + confidence score

Tech used Python · Scikit-learn · Flask · Pandas · joblib · REST API · Regex

Dataset web-page-phishing.csv — cleaned, label-encoded, duplicate-removed before training

Feature Extraction import re from urllib.parse import urlparse

def extract_features(url): features = {}

features['url_length'] = len(url) features['hostname_length'] = len(urlparse(url).hostname)

features['count_dot'] = url.count('.') features['count_hyphen'] = url.count('-') features['count_at'] = url.count('@') features['count_slash'] = url.count('/') features['count_question'] = url.count('?') features['count_equal'] = url.count('=') features['count_digit'] = sum(c.isdigit() for c in url) suspicious_words = ["secure", "account", "update", "free", "verify", "login", "bank"] features['contains_suspicious_word'] = int(any(word in url.lower() for word in suspicious_words))

match_ip = re.search(r'\d+.\d+.\d+.\d+', url) features['contains_ip'] = 1 if match_ip else 0

return list(features.values()) #Models import pandas as pd import sys from sklearn.preprocessing import LabelEncoder from sklearn.model_selection import train_test_split from sklearn.ensemble import RandomForestClassifier from sklearn.metrics import accuracy_score, classification_report import joblib

path = "web-page-phishing.csv" df = pd.read_csv(path) print("Before Cleaning:", df.shape)

print("\nColumns found:", list(df.columns))

label_col = None for col in df.columns: if col.lower() in ["status", "label", "class", "phishing", "target"]: label_col = col break

if label_col is None: sys.exit("Could not find label column. Rename target column to 'phishing' / 'label' / 'status'.")

print(f"\nUsing label column: {label_col}")

df = df.dropna(subset=[label_col]) df = df.dropna() df = df.drop_duplicates()

print("\nUnique values in label before encoding:", df[label_col].unique()) if not set(df[label_col].unique()).issubset({0, 1}): le = LabelEncoder() df[label_col] = le.fit_transform(df[label_col])

df = df.rename(columns={label_col: "status"})

print("\nFinal label distribution:") print(df["status"].value_counts())

df.to_csv("cleaned_web-page-phishing.csv", index=False) print("\nCleaned dataset saved.")

print("\n=== TRAINING MODEL ===")

X = df.drop("status", axis=1) y = df["status"]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

model = RandomForestClassifier() model.fit(X_train, y_train)

y_pred = model.predict(X_test) print("\nAccuracy:", accuracy_score(y_test, y_pred)) print("\nClassification Report:\n", classification_report(y_test, y_pred))

joblib.dump(model, "phishing_model.pkl") print("\nModel saved as: phishing_model.pkl")