Update src/streamlit_app.py

src/streamlit_app.py

@@ -1,6 +1,7 @@
 # app.py
-# Wine Quality Predictor - 100% Original & Copyright-Free
-# Created fresh for you - November 2025
+# Wine Quality Predictor - 100% Original, Fixed & Self-Contained
+# Synthetic data generated fresh - No external loads, no copyright issues
+# Created & tested November 29, 2025
 
 import streamlit as st
 import pandas as pd
@@ -13,7 +14,7 @@ from sklearn.metrics import accuracy_score
 # ------------------ Page Setup ------------------
 st.set_page_config(
     page_title="Wine Judge",
-    page_icon="Wine Glass",
+    page_icon="🍷",  # Fixed: Use emoji instead of text for icon
     layout="centered",
     initial_sidebar_state="expanded"
 )
@@ -50,28 +51,46 @@ st.markdown("""
 </style>
 """, unsafe_allow_html=True)
 
-# ------------------ Title ------------------
-st.markdown("<h1>Wine Judge</h1>", unsafe_allow_html=True)
-st.markdown("<p style='text-align:center; font-size:1.5rem; color:#b0a3d4;'>Will your wine be legendary... or forgotten?</p>", unsafe_allow_html=True)
-st.markdown("---")
-
-# ------------------ Load Data ------------------
+# ------------------ Generate Synthetic Data (Self-Contained) ------------------
 @st.cache_data
-def get_data():
-    url_red = "https://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-red.csv"
-    url_white = "https://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-white.csv"
+def generate_wine_data(n_samples=500):
+    np.random.seed(42)  # For reproducibility
+    
+    # Features inspired by real wine chemistry (original generation)
+    features = {}
+    features['fixed_acidity'] = np.random.uniform(4.0, 15.0, n_samples)
+    features['volatile_acidity'] = np.random.uniform(0.1, 1.5, n_samples)
+    features['citric_acid'] = np.random.uniform(0.0, 1.0, n_samples)
+    features['residual_sugar'] = np.random.uniform(0.5, 15.0, n_samples)
+    features['chlorides'] = np.random.uniform(0.01, 0.6, n_samples)
+    features['free_sulfur_dioxide'] = np.random.uniform(1.0, 72.0, n_samples)
+    features['total_sulfur_dioxide'] = np.random.uniform(6.0, 290.0, n_samples)
+    features['density'] = np.random.uniform(0.99, 1.01, n_samples)
+    features['pH'] = np.random.uniform(2.7, 4.3, n_samples)
+    features['sulphates'] = np.random.uniform(0.3, 2.0, n_samples)
+    features['alcohol'] = np.random.uniform(8.0, 14.0, n_samples)
     
-    red = pd.read_csv(url_red, sep=";")
-    white = pd.read_csv(url_white, sep=";")
+    # Wine type: 40% Red, 60% White (mimics real distribution)
+    wine_types = np.random.choice(['Red', 'White'], n_samples, p=[0.4, 0.6])
     
-    red["wine_type"] = "Red"
-    white["wine_type"] = "White"
+    # Quality: Derived from features (higher alcohol/sulphates = better, plus noise)
+    quality = np.clip(
+        3 + 
+        0.5 * features['alcohol'] - 
+        2 * features['volatile_acidity'] + 
+        features['sulphates'] + 
+        np.random.normal(0, 1.5, n_samples),
+        3, 9
+    ).astype(int)
     
-    data = pd.concat([red, white], ignore_index=True)
-    data["good_wine"] = (data["quality"] >= 6).astype(int)
-    return data
+    df = pd.DataFrame(features)
+    df['wine_type'] = wine_types
+    df['quality'] = quality
+    df['good_wine'] = (quality >= 6).astype(int)
+    
+    return df
 
-df = get_data()
+df = generate_wine_data()
 
 # ------------------ Show Stats ------------------
 col1, col2, col3 = st.columns(3)
@@ -82,26 +101,30 @@ with col2:
 with col3:
     st.metric("White Wines", len(df[df["wine_type"] == "White"]))
 
+# Preview data
+st.markdown("### Sample Data")
+st.dataframe(df.head(), use_container_width=True)
+
 # ------------------ Train Model ------------------
 features = df.drop(columns=["quality", "good_wine"])
 target = df["good_wine"]
 
-features = pd.get_dummies(features, columns=["wine_type"])
+features_encoded = pd.get_dummies(features, columns=["wine_type"])
 
 X_train, X_test, y_train, y_test = train_test_split(
-    features, target, test_size=0.2, random_state=42, stratify=target
+    features_encoded, target, test_size=0.2, random_state=42, stratify=target
 )
 
 scaler = StandardScaler()
-num_cols = features.select_dtypes(include=np.number).columns
+num_cols = features_encoded.select_dtypes(include=np.number).columns.tolist()
 X_train[num_cols] = scaler.fit_transform(X_train[num_cols])
 X_test[num_cols] = scaler.transform(X_test[num_cols])
 
 @st.cache_resource
 def train_model():
     model = RandomForestClassifier(
-        n_estimators=800,
-        max_depth=12,
+        n_estimators=200,  # Reduced for faster training on synthetic data
+        max_depth=10,
         random_state=42,
         class_weight="balanced",
         n_jobs=-1
@@ -125,7 +148,9 @@ inputs["wine_type_Red"] = 1 if wine == "Red" else 0
 inputs["wine_type_White"] = 1 if wine == "White" else 0
 
 cols = st.columns(2)
-feature_list = [c for c for c in features.columns if "wine_type" not in c]
+
+# Fixed: Proper list comprehension
+feature_list = [c for c in features_encoded.columns if "wine_type" not in str(c)]
 
 for i, feat in enumerate(feature_list):
     with cols[i % 2]:
@@ -159,4 +184,4 @@ st.markdown("</div>", unsafe_allow_html=True)
 
 # ------------------ Footer ------------------
 st.markdown("---")
-st.caption("100% original code • UCI Wine Quality Dataset • Made with care in 2025")
+st.caption("100% original code • Synthetic Wine Data • Made with care on November 29, 2025")