diff --git "a/main.ipynb" "b/main.ipynb"
--- "a/main.ipynb"
+++ "b/main.ipynb"
@@ -2,7 +2,7 @@
"cells": [
{
"cell_type": "code",
- "execution_count": 73,
+ "execution_count": 91,
"metadata": {},
"outputs": [],
"source": [
@@ -10,15 +10,18 @@
"import numpy as np\n",
"from sklearn.ensemble import IsolationForest\n",
"from sklearn.preprocessing import LabelEncoder\n",
+ "from sklearn.neighbors import LocalOutlierFactor\n",
+ "from sklearn.svm import OneClassSVM\n",
"import seaborn as sns\n",
"import joblib\n",
"import json\n",
- "import matplotlib.pyplot as plt"
+ "import matplotlib.pyplot as plt\n",
+ "from sklearn.metrics import classification_report, accuracy_score, silhouette_score"
]
},
{
"cell_type": "code",
- "execution_count": 74,
+ "execution_count": 92,
"metadata": {},
"outputs": [],
"source": [
@@ -27,7 +30,7 @@
},
{
"cell_type": "code",
- "execution_count": 75,
+ "execution_count": 93,
"metadata": {},
"outputs": [
{
@@ -66,7 +69,7 @@
},
{
"cell_type": "code",
- "execution_count": 76,
+ "execution_count": 94,
"metadata": {},
"outputs": [],
"source": [
@@ -76,7 +79,7 @@
},
{
"cell_type": "code",
- "execution_count": 77,
+ "execution_count": 95,
"metadata": {},
"outputs": [
{
@@ -106,7 +109,7 @@
},
{
"cell_type": "code",
- "execution_count": 78,
+ "execution_count": 96,
"metadata": {},
"outputs": [
{
@@ -130,7 +133,7 @@
},
{
"cell_type": "code",
- "execution_count": 79,
+ "execution_count": 97,
"metadata": {},
"outputs": [],
"source": [
@@ -147,12 +150,12 @@
"df['Time'] = pd.to_datetime(df['Time'], format='%H:%M').dt.hour * 3600 + pd.to_datetime(df['Time'], format='%H:%M').dt.minute * 60\n",
"\n",
"# Extract the day of the week from the Date column\n",
- "df['DayOfWeek'] = df['Date'].dt.day_name() # Monday, Tuesday, etc."
+ "df['DayOfWeek'] = df['Date'].dt.day_name()"
]
},
{
"cell_type": "code",
- "execution_count": 80,
+ "execution_count": 98,
"metadata": {},
"outputs": [],
"source": [
@@ -165,7 +168,7 @@
},
{
"cell_type": "code",
- "execution_count": 81,
+ "execution_count": 99,
"metadata": {},
"outputs": [],
"source": [
@@ -175,7 +178,7 @@
},
{
"cell_type": "code",
- "execution_count": 82,
+ "execution_count": 100,
"metadata": {},
"outputs": [
{
@@ -196,7 +199,7 @@
},
{
"cell_type": "code",
- "execution_count": 83,
+ "execution_count": 101,
"metadata": {},
"outputs": [],
"source": [
@@ -208,15 +211,14 @@
" \"Thursday\": 3, \"Friday\": 4, \"Saturday\": 5, \"Sunday\": 6\n",
"}\n",
"\n",
- "# Apply mappings to the respective columns\n",
"df[\"TransactionType\"] = df[\"TransactionType\"].map(transaction_type_mapping)\n",
"df[\"Channel\"] = df[\"Channel\"].map(channel_mapping)\n",
- "df[\"DayOfWeek\"] = df[\"DayOfWeek\"].map(day_of_week_mapping)\n"
+ "df[\"DayOfWeek\"] = df[\"DayOfWeek\"].map(day_of_week_mapping)"
]
},
{
"cell_type": "code",
- "execution_count": 84,
+ "execution_count": 102,
"metadata": {},
"outputs": [
{
@@ -240,286 +242,294 @@
},
{
"cell_type": "code",
- "execution_count": 85,
+ "execution_count": 103,
"metadata": {},
"outputs": [],
"source": [
- "# 🟢 Step 3: Select Features for Anomaly Detection\n",
- "features = [\"TransactionType\", \"Location\", \"Channel\", \"Time\", \"TransactionDuration\", \"LoginAttempts\", \"DayOfWeek\", \"TransactionAmount\", ]\n",
- "X = df[features]\n"
+ "# Select Features for Anomaly Detection\n",
+ "features = [\"TransactionType\", \"Location\", \"Channel\", \"Time\", \"TransactionDuration\", \"LoginAttempts\", \"DayOfWeek\", \"TransactionAmount\"]\n",
+ "X = df[features]"
]
},
{
- "cell_type": "code",
- "execution_count": 86,
+ "cell_type": "markdown",
"metadata": {},
- "outputs": [],
"source": [
- "model = IsolationForest(n_estimators=100, contamination=0.02, random_state=42)\n",
- "df[\"Anomaly\"] = model.fit_predict(X)\n"
+ "Isolation Forests"
]
},
{
"cell_type": "code",
- "execution_count": 87,
+ "execution_count": 104,
"metadata": {},
"outputs": [],
"source": [
- "df[\"Anomaly\"] = df[\"Anomaly\"].map({1: \"Normal\", -1: \"Anomalous\"})"
+ "# Isolation Forest Model\n",
+ "iso_forest = IsolationForest(n_estimators=100, contamination=0.02, random_state=42)\n",
+ "df[\"Anomaly_IF\"] = iso_forest.fit_predict(X)\n",
+ "df[\"Anomaly_IF\"] = df[\"Anomaly_IF\"].map({1: \"Normal\", -1: \"Anomalous\"})"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "Local Outlier Factor"
]
},
{
"cell_type": "code",
- "execution_count": 88,
+ "execution_count": 105,
"metadata": {},
"outputs": [],
"source": [
- "df.to_csv(\"cleaned_transactions.csv\", index=False)"
+ "# Local Outlier Factor Model\n",
+ "lof = LocalOutlierFactor(n_neighbors=20, contamination=0.02)\n",
+ "lof.fit(X)\n",
+ "df[\"Anomaly_LOF\"] = lof.fit_predict(X)\n",
+ "df[\"Anomaly_LOF\"] = df[\"Anomaly_LOF\"].map({1: \"Normal\", -1: \"Anomalous\"})"
]
},
{
"cell_type": "code",
- "execution_count": 89,
+ "execution_count": 106,
"metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "Model saved as isolation_forest_model.pkl\n"
- ]
- }
- ],
+ "outputs": [],
"source": [
- "# Save the trained Isolation Forest model\n",
- "joblib.dump(model, \"isolation_forest_model.pkl\")\n",
- "\n",
- "print(\"Model saved as isolation_forest_model.pkl\")"
+ "# Compute threshold for anomaly detection\n",
+ "lof_scores = -lof.negative_outlier_factor_\n",
+ "lof_threshold = np.percentile(lof_scores, 98) # Adjust percentile based on contamination level\n"
]
},
{
"cell_type": "code",
- "execution_count": 90,
+ "execution_count": 107,
"metadata": {},
"outputs": [
{
"data": {
- "text/html": [
- "\n",
- "\n",
- "
\n",
- " \n",
- " \n",
- " | \n",
- " DayOfWeek | \n",
- " Time | \n",
- " Location | \n",
- " Channel | \n",
- " TransactionDuration | \n",
- " TransactionAmount | \n",
- " Anomaly | \n",
- "
\n",
- " \n",
- " \n",
- " \n",
- " | 0 | \n",
- " 1 | \n",
- " 59340 | \n",
- " 36 | \n",
- " 0 | \n",
- " 81 | \n",
- " 14.09 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 1 | \n",
- " 1 | \n",
- " 60240 | \n",
- " 15 | \n",
- " 0 | \n",
- " 141 | \n",
- " 376.24 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 2 | \n",
- " 0 | \n",
- " 65760 | \n",
- " 23 | \n",
- " 1 | \n",
- " 56 | \n",
- " 126.29 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 3 | \n",
- " 4 | \n",
- " 59520 | \n",
- " 33 | \n",
- " 1 | \n",
- " 25 | \n",
- " 184.50 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 4 | \n",
- " 0 | \n",
- " 64260 | \n",
- " 1 | \n",
- " 1 | \n",
- " 198 | \n",
- " 13.45 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 5 | \n",
- " 0 | \n",
- " 62100 | \n",
- " 28 | \n",
- " 0 | \n",
- " 172 | \n",
- " 92.15 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 6 | \n",
- " 2 | \n",
- " 59760 | \n",
- " 39 | \n",
- " 0 | \n",
- " 139 | \n",
- " 7.08 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 7 | \n",
- " 0 | \n",
- " 64020 | \n",
- " 16 | \n",
- " 2 | \n",
- " 291 | \n",
- " 171.42 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 8 | \n",
- " 1 | \n",
- " 61140 | \n",
- " 11 | \n",
- " 2 | \n",
- " 86 | \n",
- " 106.23 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- " | 9 | \n",
- " 4 | \n",
- " 57960 | \n",
- " 26 | \n",
- " 0 | \n",
- " 120 | \n",
- " 815.96 | \n",
- " Normal | \n",
- "
\n",
- " \n",
- "
\n",
- "