File size: 14,953 Bytes
a12ca3f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
69b6195
 
 
 
 
 
 
 
 
 
 
a12ca3f
 
69b6195
a12ca3f
69b6195
a12ca3f
69b6195
 
a12ca3f
 
 
69b6195
a12ca3f
 
 
69b6195
a12ca3f
 
 
 
 
69b6195
a12ca3f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
008bc3e
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
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
import streamlit as st
import pandas as pd
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch_geometric.nn import GATConv
from torch_geometric.data import Data
import os

# Define FraudGNN class
class FraudGNN(nn.Module):
    def __init__(self, input_dim, hidden_dim, output_dim):
        super(FraudGNN, self).__init__()
        self.conv1 = GATConv(input_dim, hidden_dim, heads=4, dropout=0.3)
        self.conv2 = GATConv(hidden_dim * 4, hidden_dim, heads=1, dropout=0.3)
        self.fc = nn.Linear(hidden_dim, output_dim)

    def forward(self, data):
        x, edge_index = data.x, data.edge_index
        x = F.relu(self.conv1(x, edge_index))
        x = F.dropout(x, p=0.3, training=self.training)
        x = F.relu(self.conv2(x, edge_index))
        x = self.fc(x)
        return torch.sigmoid(x).squeeze()

# Device configuration
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

# Load model and threshold
try:
    # Try root directory first (Hugging Face Spaces working directory)
    model_path = 'fraud_gnn_model.pth'
    threshold_path = 'optimal_threshold.txt'

    # Fallback: Try relative to src/ (if files are misplaced)
    if not os.path.exists(model_path):
        model_path = os.path.join(os.path.dirname(__file__), 'fraud_gnn_model.pth')
    if not os.path.exists(threshold_path):
        threshold_path = os.path.join(os.path.dirname(__file__), 'optimal_threshold.txt')

    # Alternative: If files are in a 'models/' folder (uncomment if applicable)
    # model_path = os.path.join(os.path.dirname(__file__), '..', 'models', 'fraud_gnn_model.pth')
    # threshold_path = os.path.join(os.path.dirname(__file__), '..', 'models', 'optimal_threshold.txt')

    if not os.path.exists(model_path):
        raise FileNotFoundError(f"Model file not found at {model_path}. Please upload fraud_gnn_model.pth to the repository root.")
    if not os.path.exists(threshold_path):
        raise FileNotFoundError(f"Threshold file not found at {threshold_path}. Please upload optimal_threshold.txt to the repository root.")

    model = FraudGNN(input_dim=7, hidden_dim=16, output_dim=1).to(device)
    model.load_state_dict(torch.load(model_path, map_location=device))
    model.eval()

    with open(threshold_path, 'r') as f:
        threshold = float(f.read())
except FileNotFoundError as e:
    st.error(f"Error: {e}")
    st.stop()
except Exception as e:
    st.error(f"Error loading model or threshold: {e}")
    st.stop()

# City and state mappings
city_mapping = {
    'Atlanta': 0, 'Bronx': 1, 'Brooklyn': 2, 'Chicago': 3, 'Dallas': 4, 'Houston': 5,
    'Indianapolis': 6, 'Las Vegas': 7, 'Los Angeles': 8, 'Louisville': 9, 'Miami': 10,
    'Minneapolis': 11, 'New York': 12, 'ONLINE': 13, 'Orlando': 14, 'Philadelphia': 15,
    'San Antonio': 16, 'San Diego': 17, 'San Francisco': 18, 'Tucson': 19, 'other': 20
}
state_mapping = {
    'AK': 0, 'AL': 1, 'AR': 2, 'AZ': 3, 'Algeria': 4, 'Antigua and Barbuda': 5, 'Argentina': 6,
    'Aruba': 7, 'Australia': 8, 'Austria': 9, 'Azerbaijan': 10, 'Bahrain': 11, 'Bangladesh': 12,
    'Barbados': 13, 'Belarus': 14, 'Belgium': 15, 'Belize': 16, 'Bosnia and Herzegovina': 17,
    'Brazil': 18, 'CA': 19, 'CO': 20, 'CT': 21, 'Cabo Verde': 22, 'Cambodia': 23, 'Canada': 24,
    'Central African Republic': 25, 'Chile': 26, 'China': 27, 'Colombia': 28, 'Costa Rica': 29,
    "Cote d'Ivoire": 30, 'Croatia': 31, 'Czech Republic': 32, 'DC': 33, 'DE': 34, 'Denmark': 35,
    'Dominica': 36, 'Dominican Republic': 37, 'East Timor (Timor-Leste)': 38, 'Ecuador': 39,
    'Egypt': 40, 'Eritrea': 41, 'Estonia': 42, 'FL': 43, 'Fiji': 44, 'Finland': 45, 'France': 46,
    'GA': 47, 'Georgia': 48, 'Germany': 49, 'Ghana': 50, 'Greece': 51, 'Guatemala': 52,
    'Guyana': 53, 'HI': 54, 'Haiti': 55, 'Honduras': 56, 'Hong Kong': 57, 'Hungary': 58,
    'IA': 59, 'ID': 60, 'IL': 61, 'IN': 62, 'Iceland': 63, 'India': 64, 'Indonesia': 65,
    'Ireland': 66, 'Israel': 67, 'Italy': 68, 'Jamaica': 69, 'Japan': 70, 'Jordan': 71,
    'KS': 72, 'KY': 73, 'Kenya': 74, 'Kosovo': 75, 'Kuwait': 76, 'LA': 77, 'Latvia': 78,
    'Lebanon': 79, 'Liberia': 80, 'Lithuania': 81, 'Luxembourg': 82, 'MA': 83, 'MD': 84,
    'ME': 85, 'MI': 86, 'MN': 87, 'MO': 88, 'MS': 89, 'MT': 90, 'Macedonia': 91,
    'Malaysia': 92, 'Malta': 93, 'Mexico': 94, 'Moldova': 95, 'Monaco': 96, 'Morocco': 97,
    'Mozambique': 98, 'Myanmar (Burma)': 99, 'NC': 100, 'ND': 101, 'NE': 102, 'NH': 103,
    'NJ': 104, 'NM': 105, 'NV': 106, 'NY': 107, 'Nauru': 108, 'Netherlands': 109,
    'New Zealand': 110, 'Nicaragua': 111, 'Niger': 112, 'Nigeria': 113, 'Norway': 114,
    'OH': 115, 'OK': 116, 'OR': 117, 'Oman': 118, 'PA': 119, 'Pakistan': 120, 'Panama': 121,
    'Peru': 122, 'Philippines': 123, 'Poland': 124, 'Portugal': 125, 'RI': 126, 'Romania': 127,
    'Russia': 128, 'SC': 129, 'SD': 130, 'Saudi Arabia': 131, 'Senegal': 132, 'Serbia': 133,
    'Seychelles': 134, 'Singapore': 135, 'Slovakia': 136, 'Slovenia': 137, 'Somalia': 138,
    'South Africa': 139, 'South Korea': 140, 'Spain': 141, 'Sri Lanka': 142, 'Sudan': 143,
    'Suriname': 144, 'Sweden': 145, 'Switzerland': 146, 'Syria': 147, 'TN': 148, 'TX': 149,
    'Taiwan': 150, 'Thailand': 151, 'The Bahamas': 152, 'Tunisia': 153, 'Turkey': 154,
    'Tuvalu': 155, 'UT': 156, 'Uganda': 157, 'Ukraine': 158, 'United Arab Emirates': 159,
    'United Kingdom': 160, 'Uruguay': 161, 'Uzbekistan': 162, 'VA': 163, 'VT': 164,
    'Vatican City': 165, 'Vietnam': 166, 'WA': 167, 'WI': 168, 'WV': 169, 'WY': 170,
    'Yemen': 171, 'Zimbabwe': 172
}

def predict_fraud(transactions):
    try:
        df = pd.DataFrame(transactions, columns=[
            'Zipcode', 'Merchant_State_Code', 'User_Frequency_Per_Day',
            'Time_Difference_Hours', 'Merchant_Category_Code',
            'Merchant_City_Code', 'Transaction_Amount'
        ])
        node_features = torch.tensor(df.values, dtype=torch.float).to(device)
        
        edge_index = torch.empty((2, 0), dtype=torch.long).to(device)
        
        if len(df) > 1:
            zipcodes = node_features[:, 0].cpu().numpy()
            edge_list = []
            zipcode_threshold = 1000
            for i in range(len(df)):
                for j in range(i + 1, len(df)):
                    if abs(zipcodes[i] - zipcodes[j]) < zipcode_threshold:
                        edge_list.append([i, j])
                        edge_list.append([j, i])
            if edge_list:
                edge_index = torch.tensor(edge_list, dtype=torch.long).t().contiguous().to(device)
        
        graph_data = Data(x=node_features, edge_index=edge_index).to(device)
        
        if model is None:
            raise ValueError("Model not loaded. Check if fraud_gnn_model.pth exists.")
        
        with torch.no_grad():
            out = model(graph_data)
            out = torch.atleast_1d(out)
            pred_binary = (out > threshold).float().cpu().numpy()
            pred_proba = out.cpu().numpy()
            pred_binary = np.atleast_1d(pred_binary)
            pred_proba = np.atleast_1d(pred_proba)
        
        return pred_binary, pred_proba
    except Exception as e:
        st.error(f"Error in predict_fraud: {e}")
        return None, None

# Custom CSS for highly compact, eye-catching design
st.markdown("""
    <style>
    @import url('https://fonts.googleapis.com/css2?family=Poppins:wght@400;600&display=swap');
    @import url('https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.4.0/css/all.min.css');

    @keyframes glow {
        0% { box-shadow: 0 0 5px rgba(52, 152, 219, 0.5); }
        50% { box-shadow: 0 0 15px rgba(52, 152, 219, 0.8); }
        100% { box-shadow: 0 0 5px rgba(52, 152, 219, 0.5); }
    }
    @keyframes icon-pulse {
        0% { transform: scale(1); }
        50% { transform: scale(1.1); }
        100% { transform: scale(1); }
    }

    .stApp {
        background: #ffffff;
        max-width: 400px;
        margin: 10px auto;
        padding: 10px;
        font-family: 'Poppins', sans-serif;
        border-radius: 10px;
        box-shadow: 0 6px 20px rgba(0, 0, 0, 0.1);
        border: 2px solid transparent;
        animation: glow 3s infinite;
    }
    /* Alternative Pastel Gradient Design (uncomment to use) */
    /*
    .stApp {
        background: linear-gradient(135deg, #e6f0fa, #f3e5f5);
        max-width: 400px;
        margin: 10px auto;
        padding: 10px;
        font-family: 'Poppins', sans-serif;
        border-radius: 10px;
        box-shadow: 0 6px 20px rgba(0, 0, 0, 0.1);
        border: 2px solid transparent;
        animation: glow 3s infinite;
    }
    */
    .stTextInput > div > div > input, .stNumberInput > div > div > input, .stSelectbox > div > div > select {
        padding: 5px;
        border: 1px solid #ddd;
        border-radius: 5px;
        font-size: 0.8rem;
        background: #f9f9f9;
        transition: border-color 0.3s, box-shadow 0.3s;
    }
    .stTextInput > div > div > input:focus, .stNumberInput > div > div > input:focus, .stSelectbox > div > div > select:focus {
        outline: none;
        border-color: #3498db;
        box-shadow: 0 0 6px rgba(52, 152, 219, 0.7);
    }
    .stSelectbox > div > div > select {
        appearance: none;
        background: #f9f9f9 url('data:image/svg+xml;utf8,<svg xmlns="http://www.w3.org/2000/svg" width="10" height="10" viewBox="0 0 24 24"><path fill="%23333" d="M7 10l5 5 5-5z"/></svg>') no-repeat right 8px center;
    }
    .stButton > button {
        padding: 6px;
        background: linear-gradient(45deg, #3498db, #ff6f61);
        color: white;
        border: none;
        border-radius: 5px;
        font-size: 0.85rem;
        font-weight: 600;
        width: 100%;
        transition: transform 0.2s, box-shadow 0.3s;
    }
    .stButton > button:hover {
        transform: translateY(-2px);
        box-shadow: 0 4px 12px rgba(255, 111, 97, 0.5);
    }
    .stButton > button:active {
        transform: translateY(0);
    }
    .result-box {
        background: #f1f3f5;
        padding: 8px;
        border-radius: 6px;
        text-align: center;
        margin-top: 8px;
        border: 1px solid #ddd;
        animation: glow 3s infinite;
    }
    .result-box h2 {
        font-size: 1rem;
        color: #2c3e50;
        margin-bottom: 4px;
    }
    .result-box p {
        font-size: 0.8rem;
        color: #7f8c8d;
    }
    .fa-shield-alt {
        animation: icon-pulse 2s infinite;
    }
    .form-label {
        font-weight: 600;
        font-size: 0.75rem;
        color: #2c3e50;
        margin-bottom: 3px;
        display: flex;
        align-items: center;
    }
    .form-label i {
        color: #ff6f61;
        margin-right: 5px;
        transition: color 0.3s;
    }
    .form-label i:hover {
        color: #3498db;
    }
    .stForm {
        display: flex;
        flex-direction: column;
        gap: 6px;
    }
    </style>
""", unsafe_allow_html=True)

# Streamlit UI
st.markdown("""
    <h1 style='text-align: center; color: #2c3e50; font-size: 1.5rem; margin-bottom: 8px;'>
        <i class='fas fa-shield-alt' style='color: #ff6f61; margin-right: 8px;'></i>
        FraudShield
    </h1>
    <p style='text-align: center; font-size: 0.8rem; color: #555; margin-bottom: 8px; line-height: 1.4;'>
        Enter transaction details to detect fraud. Provide accurate zip code, merchant details, and amount.
    </p>
""", unsafe_allow_html=True)

with st.form(key="fraud_form"):
    st.markdown("<div class='form-label'><i class='fas fa-map-marker-alt'></i>Zipcode</div>", unsafe_allow_html=True)
    zipcode = st.number_input("", value=91750.0, step=0.01, format="%.2f", key="zipcode")
    
    st.markdown("<div class='form-label'><i class='fas fa-globe'></i>Merchant State</div>", unsafe_allow_html=True)
    merchant_state = st.selectbox("", sorted(state_mapping.keys()), index=sorted(state_mapping.keys()).index("TX"), key="state")
    
    st.markdown("<div class='form-label'><i class='fas fa-user-clock'></i>User Frequency Per Day</div>", unsafe_allow_html=True)
    user_freq = st.number_input("", value=1.0, step=0.01, format="%.2f", key="freq")
    
    st.markdown("<div class='form-label'><i class='fas fa-hourglass-half'></i>Time Difference (Hours)</div>", unsafe_allow_html=True)
    time_diff = st.number_input("", value=16601.95, step=0.01, format="%.2f", key="time")
    
    st.markdown("<div class='form-label'><i class='fas fa-store'></i>Merchant Category Code</div>", unsafe_allow_html=True)
    merchant_category = st.number_input("", value=5912.0, step=0.01, format="%.2f", key="category")
    
    st.markdown("<div class='form-label'><i class='fas fa-city'></i>Merchant City</div>", unsafe_allow_html=True)
    merchant_city = st.selectbox("", sorted(city_mapping.keys()), index=sorted(city_mapping.keys()).index("Houston"), key="city")
    
    st.markdown("<div class='form-label'><i class='fas fa-dollar-sign'></i>Transaction Amount</div>", unsafe_allow_html=True)
    transaction_amount = st.number_input("", value=128.35, step=0.01, format="%.2f", key="amount")
    
    submit_button = st.form_submit_button("Predict Fraud", use_container_width=True)

    if submit_button:
        try:
            if not all([zipcode, user_freq, time_diff, merchant_category, transaction_amount]):
                st.error("All fields are required.")
            elif merchant_state not in state_mapping:
                st.error(f"Invalid Merchant State: {merchant_state}")
            elif merchant_city not in city_mapping:
                st.error(f"Invalid Merchant City: {merchant_city}")
            else:
                transaction = {
                    'Zipcode': float(zipcode),
                    'Merchant_State_Code': int(state_mapping[merchant_state]),
                    'User_Frequency_Per_Day': float(user_freq),
                    'Time_Difference_Hours': float(time_diff),
                    'Merchant_Category_Code': float(merchant_category),
                    'Merchant_City_Code': int(city_mapping[merchant_city]),
                    'Transaction_Amount': float(transaction_amount)
                }
                transactions = [list(transaction.values())]
                predictions, probabilities = predict_fraud(transactions)
                
                if predictions is None or probabilities is None:
                    st.error("Prediction failed. Check server logs for details.")
                else:
                    result = 'Fraud' if predictions[0] == 1 else 'Not Fraud'
                    st.markdown(f"""
                        <div class='result-box'>
                            <h2>Transaction: {result}</h2>
                            <p>Probability of Fraud: {probabilities[0]:.4f}</p>
                        </div>
                    """, unsafe_allow_html=True)
        except Exception as e:
            st.error(f"Error: Invalid input - {str(e)}")