Upload app.py

app.py

@@ -18,48 +18,46 @@ def load_data_and_model():
         return None, None
     df = pd.read_csv(df_path)
     
-    # Standardize market value column name EARLY and DEDUPLICATE
-    for col in ['market_value_in_eur_x', 'market_value_in_eur_y', 'market_value_x', 'market_value_y']:
-        if col in df.columns:
-            df.rename(columns={col: 'market_value_in_eur'}, inplace=True)
+    # Standardize ALL possible market value names to match model
+    # The common ones are market_value_in_eur, market_value_in_eur_x, market_value_in_eur_y
+    mv_rename_map = {}
+    for col in df.columns:
+        if 'market_value' in col.lower() and 'eur' in col.lower():
+            mv_rename_map[col] = 'market_value_in_eur'
+    if mv_rename_map:
+        df.rename(columns=mv_rename_map, inplace=True)
     
-    # CRITICAL: Drop any duplicate columns created by renaming
+    # Deduplicate early
     df = df.loc[:, ~df.columns.duplicated()].copy()
     
     # Load model
     model = xgb.XGBRegressor()
     try:
-        if os.path.exists("fairvalue_xgboost.json"):
-            model.load_model("fairvalue_xgboost.json")
-        elif os.path.exists("FairValue_xgboost.json"):
-            model.load_model("FairValue_xgboost.json")
+        path = "fairvalue_xgboost.json"
+        if not os.path.exists(path):
+            path = "FairValue_xgboost.json"
+            
+        if os.path.exists(path):
+            model.load_model(path)
         else:
-            return df, "MODEL_NOT_FOUND"
+            return df, f"MODEL_FILE_NOT_FOUND (Checked: {os.getcwd()})"
     except Exception as e:
-        return df, f"ERROR: {str(e)}"
+        return df, f"LOAD_ERROR: {str(e)}"
         
     return df, model
 
 df, model = load_data_and_model()
 
 if isinstance(model, str):
-    st.error(f"Model Error: {model}. Please ensure 'fairvalue_xgboost.json' is uploaded to the root of your Hugging Face space.")
+    st.error(f"⚠️ {model}")
+    st.info("Ensure the model file is uploaded to the root of your Space.")
     st.stop()
 if df is None:
-    st.error("Data file 'data/processed/app_features.csv' not found. Please upload the 'data/' folder.")
+    st.error("⚠️ Data file not found at 'data/processed/app_features.csv'")
     st.stop()
 
-# === Data Sanity Gate ===
-if 'Age' in df.columns:
-    df['Age'] = df['Age'].clip(15, 45)
-if 'Contract_Years_Left' in df.columns:
-    df['Contract_Years_Left'] = df['Contract_Years_Left'].clip(0.5, 6.0)
-if 'market_value_in_eur' in df.columns:
-    df['market_value_in_eur'] = df['market_value_in_eur'].clip(0, 300_000_000)
-
 # ======= Sidebar =======
 st.sidebar.header("Player Transfer Profile")
-
 input_mode = st.sidebar.radio("Input Mode", ["Select Existing Player", "Create Custom Player"])
 
 name_col = next((c for c in ['name', 'name_x', 'Player_Name', 'Name'] if c in df.columns), None)
@@ -68,21 +66,19 @@ if input_mode == "Select Existing Player":
     if name_col:
         player_list = sorted(df[name_col].unique().tolist())
         selected_name = st.sidebar.selectbox("Target Database Player", player_list)
-        # Using [0:1] to keep it as a DataFrame
         player_data = df[df[name_col] == selected_name].iloc[0:1].copy()
     else:
-        st.sidebar.error("Player names column not found in CSV.")
+        st.sidebar.error("Player names column not found.")
         st.stop()
         
     st.sidebar.markdown("---")
-    st.sidebar.markdown("**Simulation Parameters** (Overrides)")
+    st.sidebar.markdown("**Simulation Parameters**")
     
-    # Use .iloc[0] to avoid KeyError: 0 when the original index is preserved
     contract_default = max(0.5, min(6.0, float(player_data['Contract_Years_Left'].iloc[0])))
     age_default = max(16, min(40, int(player_data['Age'].iloc[0])))
     
     injury_col = next((c for c in ['Injury_Days_Total_24m', 'Injury_Days'] if c in player_data.columns), None)
-    injury_default = 10
+    injury_default = 0
     if injury_col:
         injury_default = max(0, min(500, int(player_data[injury_col].iloc[0])))
     
@@ -93,34 +89,41 @@ if input_mode == "Select Existing Player":
 else:
     st.sidebar.markdown("---")
     st.sidebar.markdown("**Custom Player Attributes**")
+    # MEDIAN fallback to ensure all 20+ columns are filled
     player_data = df.median(numeric_only=True).to_frame().T
     
     contract_years = st.sidebar.slider("Contract Years Remaining", 0.5, 6.0, 2.0, 0.5)
     age = st.sidebar.slider("Age", 16, 40, 24)
     injuries = st.sidebar.number_input("Injury missed days (24m)", 0, 500, 10)
     
-    if 'market_value_in_eur' in player_data.columns:
-        m_val = st.sidebar.number_input("Current Market Value Estimation (£m)", 1.0, 150.0, 20.0)
-        player_data['market_value_in_eur'] = (m_val * 1_000_000) / 0.85
+    # Try to find the market value column to let user edit it
+    mv_col = 'market_value_in_eur'
+    if mv_col in player_data.columns:
+        m_val = st.sidebar.number_input("Current Market Value Estimation (£m)", 1.0, 200.0, 20.0)
+        player_data[mv_col] = (m_val * 1_000_000) / 0.85
 
-asking_price = st.sidebar.number_input("Selling Club Asking Price (£m)", 1.0, 200.0, 45.0)
+asking_price = st.sidebar.number_input("Selling Club Asking Price (£m)", 1.0, 300.0, 45.0)
 
 # Apply overrides
 player_data['Contract_Years_Left'] = contract_years
 player_data['Age'] = age
-if 'Injury_Days_Total_24m' in player_data.columns:
-    player_data['Injury_Days_Total_24m'] = injuries
+inj_col_to_set = next((c for c in ['Injury_Days_Total_24m', 'Injury_Days'] if c in player_data.columns), 'Injury_Days_Total_24m')
+player_data[inj_col_to_set] = injuries
 
 # Prepare inference payload
 expected_cols = model.feature_names_in_
 X_infer = player_data.reindex(columns=expected_cols, fill_value=0)
 
 if st.button("Calculate Prediction", type="primary"):
-    # Predict
+    # Debug: Check if the market value is zero
+    if X_infer['market_value_in_eur'].iloc[0] == 0:
+        st.warning("⚠️ Market Value feature is 0.0! The model will underestimate. Checking input alignment...")
+        st.write("Columns available in player_data:", player_data.columns.tolist())
+
     log_pv = model.predict(X_infer)[0]
     baseline_pv = np.expm1(log_pv)
 
-    # Apply Hard Cap logic
+    # Hard Cap logic
     conservative_bound = baseline_pv * 0.85 
     risk_discount_pct = 0.0
     if contract_years < 1.5: risk_discount_pct += 0.20
@@ -142,11 +145,10 @@ if st.button("Calculate Prediction", type="primary"):
             delta={'reference': hard_cap_m, 'increasing': {'color': "red"}, 'decreasing': {'color': "green"}},
             title={'text': "Asking Price vs Calculated Hard Cap (£m)"},
             gauge={
-                'axis': {'range': [None, max(asking_price, hard_cap_m) * 1.5]},
+                'axis': {'range': [0, max(asking_price, hard_cap_m) * 1.5]},
                 'steps': [
                     {'range': [0, hard_cap_m], 'color': "rgba(0, 255, 0, 0.3)"},
                     {'range': [hard_cap_m, hard_cap_m*1.2], 'color': "rgba(255, 255, 0, 0.3)"},
-                    {'range': [hard_cap_m*1.2, max(asking_price, hard_cap_m) * 1.5], 'color': "rgba(255, 0, 0, 0.3)"}
                 ],
                 'threshold': {'line': {'color': "white", 'width': 4}, 'thickness': 0.75, 'value': hard_cap_m}
             }
@@ -157,16 +159,20 @@ if st.button("Calculate Prediction", type="primary"):
         st.subheader("Metrics Breakdown")
         st.metric("Predicted Market Value (ML Baseline)", f"£{baseline_pv_m:.1f}m")
         if risk_discount_pct > 0:
-            st.metric("Risk Penalty Applied (Busquets Factor)", f"-£{discount_val/1_000_000:.1f}m", f"-{risk_discount_pct*100:.0f}% Risk penalty!")
+            st.metric("Risk Penalty Applied", f"-£{discount_val/1_000_000:.1f}m", f"-{risk_discount_pct*100:.0f}% Risk penalty!")
         else:
             st.metric("Risk Penalty Applied", f"£0m", "0% Risk")
         st.metric("Recommended Hard Cap", f"£{hard_cap_m:.1f}m")
         
         if asking_price > hard_cap_m:
-            st.error(f"**Winner's Curse Risk:** The asking price is £{asking_price - hard_cap_m:.1f}m OVER the Cap.")
+            st.error(f"Winner's Curse Risk: The asking price is £{asking_price - hard_cap_m:.1f}m OVER the Cap.")
         else:
-            st.success("The asking price is within acceptable statistical margins.")
+            st.success("Asking price is within acceptable margins.")
 
+    # SHAP / Diagnostics
+    with st.expander("Show Model Input Diagnostics"):
+        st.write("X_infer Payload:", X_infer)
+    
     st.subheader("Algorithmic Interpretability (SHAP)")
     try:
         explainer = shap.TreeExplainer(model)
@@ -181,6 +187,6 @@ if st.button("Calculate Prediction", type="primary"):
 st.markdown("---")
 st.subheader("Model Performance Assessment")
 st.markdown(f"**Training Set Size:** `{len(df):,}` verified transfers")
-st.markdown(f"**Features Engined:** `{len(model.feature_names_in_)}` columns")
+st.markdown(f"**Features Engined:** `{len(expected_cols)}` columns")
 st.markdown("**Validation MAE (Mean Absolute Error):** `~£23,980,000` *(Statistical Range)*")
 st.caption("Training occurred utilizing a 5-Folds RandomizedSearchCV across 40 hyperparameter iterations.")