Upload 6 files

app.py

@@ -0,0 +1,85 @@
+import gradio as gr
+import pandas as pd
+import joblib
+
+# =========================
+# LOAD MODEL ARTIFACTS
+# =========================
+rf_model = joblib.load("hf_model_files/rf_fraud_model.joblib")
+scaler = joblib.load("hf_model_files/scaler.joblib")
+encoders = joblib.load("hf_model_files/label_encoders.joblib")
+
+# =========================
+# EXTRACT VALID CATEGORIES
+# =========================
+transaction_types = list(encoders['transaction_type'].classes_)
+channels = list(encoders['channel'].classes_)
+locations = list(encoders['location'].classes_)
+
+
+# =========================
+# PREDICTION FUNCTION
+# =========================
+def predict_fraud(transaction_type, channel, location, amount, old_balance, new_balance):
+    try:
+        # Create input dataframe
+        input_df = pd.DataFrame([{
+            "transaction_type": transaction_type,
+            "channel": channel,
+            "location": location,
+            "amount": float(amount),
+            "old_balance": float(old_balance),
+            "new_balance": float(new_balance)
+        }])
+
+        # Encode categorical features
+        for col in ['transaction_type', 'channel', 'location']:
+            input_df[col] = encoders[col].transform(input_df[col])
+
+        # Scale features
+        X_scaled = scaler.transform(input_df)
+
+        # Predict
+        pred = rf_model.predict(X_scaled)[0]
+        prob = rf_model.predict_proba(X_scaled)[0][1]
+
+        return (
+            "🚨 Fraud Detected" if pred == 1 else "✅ Not Fraud",
+            f"{prob:.4f}"
+        )
+
+    except Exception as e:
+        return "Error", str(e)
+
+
+# =========================
+# GRADIO INTERFACE
+# =========================
+iface = gr.Interface(
+    fn=predict_fraud,
+    inputs=[
+        gr.Dropdown(transaction_types, label="Transaction Type"),
+        gr.Dropdown(channels, label="Channel"),
+        gr.Dropdown(locations, label="Location"),
+        gr.Number(label="Amount"),
+        gr.Number(label="Old Balance"),
+        gr.Number(label="New Balance"),
+    ],
+    outputs=[
+        gr.Textbox(label="Prediction"),
+        gr.Textbox(label="Fraud Probability")
+    ],
+    title="💳 Fraud Detection System (South Africa)",
+    description="Select transaction details to predict whether a transaction is fraudulent.",
+
+    examples=[
+        [transaction_types[0], channels[0], locations[0], 5000, 10000, 5000],
+        [transaction_types[-1], channels[-1], locations[-1], 200, 500, 300],
+    ]
+)
+
+# =========================
+# RUN APP
+# =========================
+if __name__ == "__main__":
+    iface.launch()

hf_model_files/label_encoders.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b97f934e8112187d17faeba8d3459c0118165210eb7c7e6b6e420347cd00189b
+size 1262

hf_model_files/rf_fraud_model.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99c9b8a2824d6e77c1c47ebdfe3e339e1318d0165852fd4ff3dace0b1ee01dd1
+size 50247033

hf_model_files/scaler.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c14022688f5904a4f574541fb3c0e96badec120e2843e52b826f655689bf9792
+size 1103

requirements.txt.txt

@@ -0,0 +1,6 @@
+pandas
+numpy
+scikit-learn
+imbalanced-learn
+joblib
+gradio