Upload 3 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+model.ubj filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,74 @@
+import streamlit as st
+import pandas as pd
+import xgboost as xgb
+import os
+import pickle
+
+# Set page configuration
+st.set_page_config(page_title="Podcast Prediction", layout="centered")
+
+# --- 1. LOAD ASSETS ---
+@st.cache_resource
+def load_assets():
+    current_dir = os.path.dirname(os.path.abspath(__file__))
+    model_path = os.path.join(current_dir, 'model.ubj')
+    columns_path = os.path.join(current_dir, 'model_columns.pkl')
+
+    # Load XGBoost model
+    model = xgb.XGBRegressor()
+    model.load_model(model_path)
+    
+    # Load the column names list
+    with open(columns_path, 'rb') as f:
+        model_columns = pickle.load(f)
+    
+    return model, model_columns
+
+try:
+    model, model_columns = load_assets()
+    st.success("Model and Column definitions loaded!")
+except Exception as e:
+    st.error(f"Error loading assets: {e}")
+    st.stop()
+
+# --- 2. UI INPUTS ---
+st.title("🎙️ Podcast Listening Time Predictor")
+
+with st.form("input_form"):
+    col1, col2 = st.columns(2)
+    with col1:
+        category = st.selectbox("Category", ["Technology", "True Crime", "Comedy", "Health", "Business"])
+        episode_length = st.number_input("Length (Mins)", value=30)
+    with col2:
+        day_of_week = st.selectbox("Day", ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"])
+        user_history = st.number_input("User History (Mins)", value=100)
+    
+    submit = st.form_submit_button("Predict")
+
+# --- 3. PREDICTION & ALIGNMENT ---
+if submit:
+    # 1. Create initial DataFrame
+    input_df = pd.DataFrame({
+        'episode_length': [episode_length],
+        'user_history': [user_history],
+        'category': [category],
+        'day_of_week': [day_of_week]
+    })
+
+    # 2. Dummy Encoding (Match the logic used in training)
+    input_df = pd.get_dummies(input_df)
+
+    # 3. ALIGNMENT: The "Magic" step to prevent errors or stuck predictions
+    # Add missing columns (that exist in model_columns but not in current input)
+    for col in model_columns:
+        if col not in input_df.columns:
+            input_df[col] = 0
+            
+    # Reorder columns to match the training order exactly
+    input_df = input_df[model_columns]
+
+    # 4. Final Prediction
+    prediction = model.predict(input_df)
+    
+    st.divider()
+    st.subheader(f"Prediction: {prediction[0]:.2f} Minutes")

model.ubj

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71de975ce86fba817157eada679132026bb6eb6bdd0ada6e86b7371f472a2c94
+size 4967249