Update app.py

app.py

@@ -1,58 +1,65 @@
+import gradio as gr
 import pandas as pd
 import joblib
-import gradio as gr
-import seaborn as sns
 import matplotlib.pyplot as plt
+import seaborn as sns
 
-# Load the trained model
+# Load trained model
 model = joblib.load("anomaly_detector_rf_model.pkl")
 
-# Define feature order
-feature_order = ['hour', 'day_of_week', 'is_weekend', 'amount', 'merchant_avg_amount', 
-                 'amount_zscore', 'log_amount', 'type_atm_withdrawal', 'type_credit', 
-                 'type_debit', 'merchant_encoded']
+# Define feature columns used during training
+feature_cols = [
+    "hour", "day_of_week", "is_weekend", "merchant_avg_amount",
+    "amount_zscore", "log_amount", "type_atm_withdrawal", "type_credit",
+    "type_debit", "merchant_encoded"
+]
 
-def detect_anomalies(data):
-    df = pd.DataFrame(data)
-    df = df[feature_order]  # Ensure correct feature order
-    df['is_anomalous'] = model.predict(df)
+# Function to detect anomalies
+def detect_anomalies(df):
+    # Ensure only the required features are used & in correct order
+    missing_features = [col for col in feature_cols if col not in df.columns]
+    extra_features = [col for col in df.columns if col not in feature_cols]
+
+    if missing_features:
+        return f"Missing features: {missing_features}"
+
+    if extra_features:
+        df = df[feature_cols]  # Select only relevant columns
+
+    # Make predictions
+    df["is_anomalous"] = model.predict(df)
     
-    # Filter anomalies and display relevant details
-    anomalies = df[df['is_anomalous'] == 1][['transaction_id', 'merchant', 'location', 'amount']]
-    return anomalies
+    # Filter anomalies
+    anomalies = df[df["is_anomalous"] == 1]
+
+    return anomalies[["transaction_id", "merchant", "location", "amount"]]
 
-# Function to generate plots
-def generate_plots(df):
+# Function to visualize anomalies
+def plot_charts(df):
     fig, axes = plt.subplots(2, 2, figsize=(12, 10))
-    
-    sns.countplot(data=df, x='is_anomalous', palette='Set2', ax=axes[0, 0])
-    axes[0, 0].set_title("Anomaly Distribution")
-    
-    sns.countplot(data=df, y='merchant', order=df['merchant'].value_counts().index, palette='viridis', ax=axes[0, 1])
-    axes[0, 1].set_title("Transactions by Merchant")
-    
-    sns.histplot(df['amount'], bins=30, kde=True, color='blue', ax=axes[1, 0])
-    axes[1, 0].set_title("Transaction Amount Distribution")
-    
-    sns.scatterplot(data=df, x='amount', y='merchant_avg_amount', hue='is_anomalous', palette='coolwarm', ax=axes[1, 1])
-    axes[1, 1].set_title("Amount vs. Merchant Average Amount")
-    
+    sns.histplot(df["amount"], bins=30, kde=True, ax=axes[0, 0])
+    sns.boxplot(x=df["amount"], ax=axes[0, 1])
+    sns.countplot(x=df["day_of_week"], ax=axes[1, 0])
+    sns.barplot(x=df["merchant"], y=df["amount"], ax=axes[1, 1])
+
     plt.tight_layout()
     return fig
 
 # Gradio Interface
-def app_interface(file):
-    df = pd.read_csv(file.name)
+def app_interface(csv_file):
+    df = pd.read_csv(csv_file)
     anomalies = detect_anomalies(df)
-    plot = generate_plots(df)
-    return anomalies, plot
+    fig = plot_charts(df)
+    
+    return anomalies, fig
 
 interface = gr.Interface(
     fn=app_interface,
-    inputs=[gr.File(label="Upload Transaction Data (CSV)")],
-    outputs=[gr.Dataframe(label="Detected Anomalies"), gr.Plot(label="Transaction Analysis Charts")],
-    title="Financial Anomaly Detection",
-    description="Upload a transaction dataset to detect financial anomalies and visualize transaction patterns."
+    inputs="file",
+    outputs=["dataframe", "plot"],
+    title="Financial Anomaly Detector",
+    description="Upload a transaction CSV file to detect fraudulent transactions."
 )
 
+# Launch the Gradio app with public access
 interface.launch(share=True)