Update app.py

app.py

@@ -4,35 +4,36 @@ import numpy as np
 import joblib
 import xgboost as xgb
 from tensorflow.keras.models import load_model
-import pickle
+from sklearn.preprocessing import MinMaxScaler
 import matplotlib.pyplot as plt
 import seaborn as sns
 
 # Load models & scalers
-xgb_clf = xgb.XGBClassifier(); xgb_clf.load_model("xgb_model.json")
+xgb_clf = xgb.XGBClassifier()
+xgb_clf.load_model("xgb_model.json")
+
 xgb_reg = joblib.load("xgb_pipeline_model.pkl")
+
 scaler_X = joblib.load("scaler_X.pkl")
 scaler_y = joblib.load("scaler_y.pkl")
+
 lstm_model = load_model("lstm_revenue_model.keras")
 
 # Prediction + Plot functions
-def classify_fn(df):
-    # df: pandas DataFrame of features
+def classify_fn(df: pd.DataFrame):
     preds = xgb_clf.predict(df)
     probs = xgb_clf.predict_proba(df)
-    # Bar chart for first prediction probabilities
     fig, ax = plt.subplots()
-    ax.bar(['No Bankruptcy','Bankruptcy'], probs[0], color=['#4CAF50','#F44336'])
-    ax.set_ylim(0,1)
+    ax.bar(['No Bankruptcy', 'Bankruptcy'], probs[0], color=['#4CAF50', '#F44336'])
+    ax.set_ylim(0, 1)
     ax.set_title('Bankruptcy Probability')
     ax.set_ylabel('Probability')
     plt.tight_layout()
     return {"Predicted Label": int(preds[0])}, fig
 
 
-def regress_fn(df):
+def regress_fn(df: pd.DataFrame):
     preds = xgb_reg.predict(df)
-    # Histogram of predictions
     fig, ax = plt.subplots()
     sns.histplot(preds, bins=20, kde=True, ax=ax)
     ax.set_title('Anomaly Score Distribution')
@@ -41,13 +42,11 @@ def regress_fn(df):
     return preds.tolist(), fig
 
 
-def lstm_fn(seq_str):
-    # seq_str: comma-separated Q0-Q9 revenues
+def lstm_fn(seq_str: str):
     vals = np.array(list(map(float, seq_str.split(',')))).reshape(1, -1)
     vals_s = scaler_X.transform(vals).reshape((1, vals.shape[1], 1))
     pred_s = lstm_model.predict(vals_s)
-    pred = scaler_y.inverse_transform(pred_s)[0,0]
-    # Plot input series + predicted point
+    pred = scaler_y.inverse_transform(pred_s)[0, 0]
     fig, ax = plt.subplots()
     ax.plot(range(10), vals.flatten(), marker='o', label='Input Revenue')
     ax.plot(10, pred, marker='X', markersize=10, color='red', label='Predicted Q10')
@@ -68,25 +67,29 @@ h1, h2 {color: #333;}
 demo = gr.Blocks(css=grid_css)
 with demo:
     gr.Markdown("# 🚀 FinSight 360™ Dashboard")
-    gr.Markdown("Comprehensive financial AI: Bankruptcy Classification, Anomaly Scoring, and Revenue Forecasting.")
+    gr.Markdown("Comprehensive financial AI:\\n- Bankruptcy Classification\\n- Anomaly Scoring\\n- Revenue Forecasting")
+
     with gr.Tab("🏦 Bankruptcy Classifier"):
-        gr.Markdown("Upload a single row of company features (CSV or paste) to predict bankruptcy:")
-        inp1 = gr.Dataframe(headers="...", datatype=["number"]*20 + ["text"]*4, label="Features DataFrame")
-        lbl1 = gr.Label(label="Predicted Label")
+        gr.Markdown("**Upload company features** (as DataFrame) to predict bankruptcy:")
+        inp1 = gr.Dataframe(type="pandas", label="Features DataFrame")
+        out1 = gr.Label(label="Predicted Label")
         plt1 = gr.Plot()
-        inp1.submit(classify_fn, inp1, [lbl1, plt1])
+        inp1.submit(classify_fn, inp1, [out1, plt1])
+
     with gr.Tab("📈 Anomaly Regression"):
-        gr.Markdown("Upload company features to predict anomaly score:")
-        inp2 = gr.Dataframe(headers="...", datatype=["number"]*100, label="Features DataFrame")
+        gr.Markdown("**Upload company features** (as DataFrame) to predict anomaly score:")
+        inp2 = gr.Dataframe(type="pandas", label="Features DataFrame")
         out2 = gr.Textbox(label="Predicted Scores List")
         plt2 = gr.Plot()
         inp2.submit(regress_fn, inp2, [out2, plt2])
+
     with gr.Tab("📊 LSTM Revenue Forecast"):
-        gr.Markdown("Enter last 10 quarterly revenues (comma-separated) to forecast Q10 revenue:")
+        gr.Markdown("**Enter last 10 quarterly revenues** (comma-separated) to forecast Q10 revenue:")
         inp3 = gr.Textbox(placeholder="e.g. 1000,1200,1100,...", label="Q0–Q9 Revenues")
         out3 = gr.Number(label="Predicted Q10 Revenue")
         plt3 = gr.Plot()
         inp3.submit(lstm_fn, inp3, [out3, plt3])
-    gr.Markdown("---\n*Built with ❤️ by FinSight AI Team*")
+
+    gr.Markdown("---\\n*SDG 9: Industry, Innovation and Infrastructure*")
 
 demo.launch()