File size: 10,596 Bytes
8a08300
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import streamlit as st
import pandas as pd
import requests
import plotly.graph_objects as ob
import json
from datetime import datetime, time
import plotly.express as px
from src.features.constants import category_names, job_names

# 1. Configuration & Layout
st.set_page_config(page_title="PayShield Monitor", layout="wide", page_icon="🛡️")

# Custom CSS for FinTech Aesthetic
st.markdown("""
    <style>
    .main {
        background-color: #f8f9fa;
    }
    .stMetric {
        background-color: #ffffff;
        padding: 15px;
        border-radius: 10px;
        box-shadow: 0 2px 4px rgba(0,0,0,0.05);
    }
    .status-banner {
        padding: 20px;
        border-radius: 10px;
        color: white;
        text-align: center;
        font-weight: bold;
        font-size: 24px;
        margin-bottom: 20px;
    }
    .banner-safe {
        background-color: #28a745;
    }
    .banner-fraud {
        background-color: #dc3545;
    }
    .banner-shadow {
        background-color: #ffc107;
        color: #212529;
    }
    </style>
    """, unsafe_allow_html=True)

st.title("🛡️ PayShield-ML Analyst Workbench")
st.markdown("---")

# 2. Sidebar: Transaction Simulator
with st.sidebar:
    st.header("🛒 Transaction Simulator")
    
    with st.expander("👤 User Profile", expanded=True):
        user_id = st.text_input("User ID", value="u12345")
        job = st.selectbox("Job Title", options=sorted(job_names), index=sorted(job_names).index("Engineer, biomedical") if "Engineer, biomedical" in job_names else 0)
        dob = st.date_input("Date of Birth", value=datetime(1985, 3, 20))
        gender = st.radio("Gender", ["M", "F"], horizontal=True)

    with st.expander("💳 Transaction Details", expanded=True):
        amt = st.number_input("Amount ($)", min_value=0.01, value=150.0, step=10.0)
        category = st.selectbox("Category", options=sorted(category_names), index=sorted(category_names).index("grocery_pos") if "grocery_pos" in category_names else 0)
        trans_date = st.date_input("Transaction Date", value=datetime.now())
        trans_time = st.time_input("Transaction Time", value=time(14, 30))

    with st.expander("📍 Location Details", expanded=True):
        st.caption("User Coordinates")
        lat = st.number_input("User Lat", value=40.7128, format="%.4f")
        long = st.number_input("User Long", value=-74.0060, format="%.4f")
        
        st.caption("Merchant Coordinates")
        merch_lat = st.number_input("Merchant Lat", value=40.7200, format="%.4f")
        merch_long = st.number_input("Merchant Long", value=-74.0100, format="%.4f")

    with st.expander("🛠️ Advanced: Feature Overrides", expanded=False):
        st.caption("Force specific values for sensitivity analysis")
        override_trans_count = st.number_input("Count (24h)", min_value=0, value=0, help="Leave 0 to use real-time data")
        override_avg_spend = st.number_input("Avg Spend (24h)", min_value=0.0, value=0.0, help="Leave 0 to use real-time data")
        override_user_avg_all_time = st.number_input("User Avg (All Time)", min_value=0.0, value=0.0, help="Leave 0 to use calculated value")
        
    analyze_btn = st.button("🚀 Analyze Transaction", use_container_width=True, type="primary")

# 3. Main Dashboard Logic
if analyze_btn:
    # Construct Payload
    trans_dt = datetime.combine(trans_date, trans_time)
    payload = {
        "user_id": user_id,
        "trans_date_trans_time": trans_dt.strftime("%Y-%m-%d %H:%M:%S"),
        "amt": amt,
        "lat": lat,
        "long": long,
        "merch_lat": merch_lat,
        "merch_long": merch_long,
        "job": job,
        "category": category,
        "gender": gender,
        "dob": dob.strftime("%Y-%m-%d")
    }
    
    # Add overrides if set
    if override_trans_count > 0:
        payload["trans_count_24h"] = override_trans_count
    if override_avg_spend > 0:
        payload["avg_spend_24h"] = override_avg_spend
    if override_user_avg_all_time > 0:
        payload["user_avg_amt_all_time"] = override_user_avg_all_time

    try:
        # Step 1: Call API
        with st.spinner("Analyzing with XGBoost Engine..."):
            # Use docker-internal DNS or localhost depending on environment
            import os
            api_url = os.getenv("API_URL", "http://127.0.0.1:8000/v1/predict")
            response = requests.post(api_url, json=payload, timeout=5)
            
        if response.status_code == 200:
            res = response.json()
            score = res["risk_score"]
            decision = res["decision"]
            latency = res["latency_ms"]
            is_shadow = res.get("shadow_mode", False)
            features = res.get("features", {}) # Get real features used

            # Columns for high-level metrics
            m1, m2, m3 = st.columns(3)
            with m1:
                st.metric("Decision Result", decision)
            with m2:
                st.metric("Risk Score", f"{score:.1f}/100")
            with m3:
                st.metric("Inference Latency", f"{latency:.2f}ms")

            # Decision Banner
            if decision == "BLOCK":
                st.markdown('<div class="status-banner banner-fraud">❌ FRAUD DETECTED - TRANSACTION BLOCKED</div>', unsafe_allow_html=True)
            else:
                st.markdown('<div class="status-banner banner-safe">✅ TRANSACTION APPROVED</div>', unsafe_allow_html=True)

            if is_shadow and decision == "BLOCK":
                 st.markdown('<div class="status-banner banner-shadow">⚠️ SHADOW MODE: Transaction allowed in simulation.</div>', unsafe_allow_html=True)

            # 4. Visualizations
            v1, v2 = st.columns([1, 1])

            with v1:
                st.subheader("🎯 Risk Gauge")
                fig = ob.Figure(ob.Indicator(
                    mode = "gauge+number",
                    value = score,
                    domain = {'x': [0, 1], 'y': [0, 1]},
                    title = {'text': "Confidence Score (%)"},
                    gauge = {
                        'axis': {'range': [0, 100]},
                        'bar': {'color': "#333"},
                        'steps': [
                            {'range': [0, 50], 'color': "rgba(40, 167, 69, 0.3)"},
                            {'range': [50, 82], 'color': "rgba(255, 193, 7, 0.3)"},
                            {'range': [82, 100], 'color': "rgba(220, 53, 69, 0.3)"}
                        ],
                        'threshold': {
                            'line': {'color': "black", 'width': 4},
                            'thickness': 0.75,
                            'value': 82
                        }
                    }
                ))
                fig.update_layout(height=350, margin=dict(l=20, r=20, t=50, b=20))
                st.plotly_chart(fig, use_container_width=True)

            with v2:
                st.subheader("📊 Feature Explainability (SHAP)")
                shap_data = res.get("shap_values", {})
                
                if shap_data:
                    # Create DataFrame from real SHAP values
                    shap_df = pd.DataFrame([
                        {"Feature": k, "Impact": v, "Abs_Impact": abs(v)}
                        for k, v in shap_data.items()
                    ]).sort_values("Abs_Impact", ascending=True)
                    
                    # Color based on positive/negative contribution
                    colors = ["#dc3545" if x > 0 else "#28a745" for x in shap_df["Impact"]]
                    
                    fig_shap = px.bar(
                        shap_df,
                        x="Impact",
                        y="Feature",
                        orientation="h",
                        title="Top Feature Contributions to Risk Score",
                        color="Impact",
                        color_continuous_scale=["#28a745", "#ffc107", "#dc3545"],
                        labels={"Impact": "SHAP Value (Impact on Prediction)"}
                    )
                    fig_shap.update_layout(
                        height=350,
                        margin=dict(l=20, r=20, t=50, b=20),
                        showlegend=False
                    )
                    st.plotly_chart(fig_shap, use_container_width=True)
                    st.caption("🔴 Red = Increases fraud risk | 🟢 Green = Decreases fraud risk")
                else:
                    st.info("SHAP explainability not available. Enable it in API settings.")

            # 5. Internal Data State (Architectural Demo)
            st.markdown("---")
            d1, d2 = st.columns([1, 1])
            
            with d1:
                st.subheader("🗄️ Feature Store Payload")
                if features:
                    st.info("Real-time features used for inference (from Redis or Overrides):")
                    # Format for better readability
                    display_features = {
                        "Velocity (24h)": features.get("trans_count_24h"),
                        "Avg Spend (24h)": f"${features.get('avg_spend_24h', 0):.2f}",
                        "Current/Avg Ratio": f"{features.get('amt_to_avg_ratio_24h', 0):.2f}x",
                        "User Avg (All Time)": f"${features.get('user_avg_amt_all_time', 0):.2f}"
                    }
                    st.table(pd.DataFrame([display_features]))
                else:
                     st.warning("No feature data returned from API.")

            with d2:
                st.subheader("🔍 API RAW Response")
                with st.expander("View JSON", expanded=False):
                    st.json(res)

        else:
            st.error(f"API Error: Status {response.status_code}")
            st.json(response.json())

    except requests.exceptions.ConnectionError:
        st.error("🛑 Connection Failed: Could not connect to Inference API. Is the server running at http://127.0.0.1:8000?")
        st.info("Try running: `uv run uvicorn src.api.main:app --reload`")
    except Exception as e:
        st.error(f"⚠️ Unexpected Error: {str(e)}")

else:
    # Landing State
    st.info("👈 Fill in the details in the sidebar and click 'Analyze Transaction' to start.")
    
    # Hero/Demo Content
    c1, c2, c3 = st.columns(3)
    c1.markdown("### ⚡ Low Latency\nSub-50ms inference utilizing XGBoost and Redis.")
    c2.markdown("### 📋 Explainable\nSHAP integration for transparent fraud scoring.")
    c3.markdown("### 🧪 Shadow Mode\nSafe production testing of new model versions.")