Update app.py

app.py

@@ -6,7 +6,7 @@ import torch
 from chronos import ChronosPipeline
 
 class TimeSeriesForecaster:
-    def __init__(self, model_name="amazon/chronos-t5-small"):
+    def __init__(self, model_name="amazon/chronos-t5-large"):
         self.pipeline = ChronosPipeline.from_pretrained(
             model_name,
             device_map="cuda" if torch.cuda.is_available() else "cpu",
@@ -15,7 +15,7 @@ class TimeSeriesForecaster:
         self.original_series = None
         self.context = None
 
-    def preprocess_data(self, df, date_column, value_column, context_length=30, prediction_length=7):
+    def preprocess_data(self, df, date_column, value_column, context_length=100, prediction_length=365):
         """
         Prepare time series data from DataFrame
         """
@@ -36,151 +36,168 @@ class TimeSeriesForecaster:
         
         return self.context, context_length
 
-    def forecast(self, context, prediction_length=7, num_samples=100):
+    def forecast(self, context, prediction_length=365, num_samples=100):
         """
         Perform time series forecasting
         """
         forecasts = self.pipeline.predict(context, prediction_length, num_samples=num_samples)
         return forecasts
 
-    def visualize_forecast(self, context, forecasts):
+    def visualize_forecast(self, forecasts, original_series):
         """
-        Create visualization of predictions
+        Create comprehensive visualization of predictions
         """
-        plt.figure(figsize=(12, 6), facecolor='#f0f2f6')
-        
-        # Plot original series
-        plt.plot(range(len(self.original_series)), self.original_series, label='Historical Data', color='#1E88E5', linewidth=2)
+        plt.figure(figsize=(16, 8), dpi=100, facecolor='white')
         
         # Calculate forecast statistics
         forecast_np = forecasts[0].numpy()
         low, median, high = np.quantile(forecast_np, [0.1, 0.5, 0.9], axis=0)
         
-        # Plot forecast
-        forecast_index = range(len(self.original_series), len(self.original_series) + len(median))
-        plt.plot(forecast_index, median, color='#D81B60', linewidth=2, label='Median Forecast')
-        plt.fill_between(forecast_index, low, high, color='#D81B60', alpha=0.3, label='80% Prediction Interval')
+        # Plot original series
+        plt.plot(
+            range(len(original_series)), 
+            original_series, 
+            label='Historical Data', 
+            color='#2C3E50', 
+            linewidth=2,
+            alpha=0.7
+        )
+        
+        # Forecast index
+        forecast_index = range(len(original_series), len(original_series) + len(median))
+        
+        # Plot median forecast
+        plt.plot(
+            forecast_index, 
+            median, 
+            color='#3498DB', 
+            linewidth=3, 
+            label='Median Forecast'
+        )
         
-        plt.title('Time Series Forecasting', fontsize=16, fontweight='bold')
-        plt.xlabel('Time Index', fontsize=12)
-        plt.ylabel('Value', fontsize=12)
-        plt.legend(frameon=True)
-        plt.grid(True, linestyle='--', alpha=0.7)
+        # Plot prediction interval
+        plt.fill_between(
+            forecast_index, 
+            low, 
+            high, 
+            color='#3498DB', 
+            alpha=0.2, 
+            label='90% Prediction Interval'
+        )
+        
+        plt.title('Advanced Time Series Forecast', fontsize=18, fontweight='bold', color='#2C3E50')
+        plt.xlabel('Time Steps', fontsize=12, color='#34495E')
+        plt.ylabel('Value', fontsize=12, color='#34495E')
+        plt.legend(frameon=False)
+        plt.grid(True, linestyle='--', color='#BDC3C7', alpha=0.5)
+        
+        # Sophisticated styling
+        plt.tight_layout()
         
         return plt
 
 def main():
-    # Set page configuration
+    # Page configuration
     st.set_page_config(
-        page_title="Time Series Forecaster", 
-        page_icon="📈", 
-        layout="wide",
-        initial_sidebar_state="collapsed"
+        page_title="Forecast Maestro", 
+        page_icon="📊", 
+        layout="wide"
     )
 
-    # Custom CSS for modern look
+    # Modern, minimalist styling
     st.markdown("""
     <style>
-    /* Modern, clean design */
     .stApp {
-        background-color: #f0f2f6;
-        font-family: 'Inter', sans-serif;
+        background-color: #FFFFFF;
+        font-family: 'Inter', 'Roboto', sans-serif;
     }
     
-    /* Card-like containers */
-    .card {
-        background-color: white;
-        border-radius: 10px;
+    .stButton>button {
+        background-color: #3498DB;
+        color: white;
+        border: none;
+        border-radius: 8px;
+        padding: 10px 20px;
+        transition: all 0.3s ease;
+        font-weight: 600;
+    }
+    
+    .stButton>button:hover {
+        background-color: #2980B9;
+        transform: scale(1.05);
         box-shadow: 0 4px 6px rgba(0,0,0,0.1);
-        padding: 20px;
-        margin-bottom: 20px;
     }
     
-    /* Stylish file uploader */
+    /* Sleek file uploader */
     .stFileUploader {
-        background-color: #f8f9fa;
-        border: 2px dashed #6c757d;
+        background-color: #F8F9FA;
+        border: 2px solid #3498DB;
         border-radius: 10px;
         padding: 20px;
         text-align: center;
-    }
-    
-    /* Buttons */
-    .stButton>button {
-        background-color: #1E88E5;
-        color: white;
-        border-radius: 6px;
         transition: all 0.3s ease;
     }
-    .stButton>button:hover {
-        background-color: #1565c0;
-        transform: scale(1.05);
+    .stFileUploader:hover {
+        border-color: #2980B9;
+        background-color: #EAF2F8;
     }
     </style>
     """, unsafe_allow_html=True)
 
-    # Title and description
-    st.markdown("<h1 style='text-align: center; color: #1E88E5;'>🕰️ Time Series Forecaster</h1>", unsafe_allow_html=True)
-    st.markdown("<p style='text-align: center; color: #6c757d;'>Predict future trends with advanced machine learning</p>", unsafe_allow_html=True)
+    # Elegant title
+    st.markdown(
+        "<h1 style='text-align: center; color: #3498DB; margin-bottom: 30px;'>🔮 Forecast Maestro</h1>", 
+        unsafe_allow_html=True
+    )
 
-    # File upload section
-    st.markdown("<div class='card'>", unsafe_allow_html=True)
+    # File upload
     uploaded_file = st.file_uploader(
-        "Upload CSV File", 
+        "Upload Time Series Data", 
         type=['csv'], 
-        help="Upload a CSV file with time series data",
-        label_visibility="collapsed"
+        help="CSV with timestamp and numeric columns"
     )
-    st.markdown("</div>", unsafe_allow_html=True)
 
-    # Data and forecast configuration
     if uploaded_file is not None:
         # Read CSV
         df = pd.read_csv(uploaded_file)
         
-        # Configuration card
-        st.markdown("<div class='card'>", unsafe_allow_html=True)
+        # Column selection
         col1, col2 = st.columns(2)
         
         with col1:
-            date_column = st.selectbox(
-                'Select Date Column', 
-                options=df.columns,
-                help="Choose the column representing timestamps"
-            )
+            date_column = st.selectbox('Date Column', options=df.columns)
         
         with col2:
             value_column = st.selectbox(
-                'Select Value Column', 
-                options=[col for col in df.columns if col != date_column],
-                help="Choose the numeric column to forecast"
+                'Value Column', 
+                options=[col for col in df.columns if col != date_column]
             )
         
-        # Prediction parameters
+        # Advanced prediction settings
         col3, col4 = st.columns(2)
         
         with col3:
             context_length = st.slider(
-                'Context Length', 
-                min_value=10, 
-                max_value=100, 
-                value=30,
-                help="Number of historical data points to use for prediction"
+                'Historical Context', 
+                min_value=30, 
+                max_value=500, 
+                value=100,
+                help="Number of past data points to analyze"
             )
         
         with col4:
             prediction_length = st.slider(
-                'Prediction Length', 
-                min_value=1, 
-                max_value=30, 
-                value=7,
+                'Forecast Horizon', 
+                min_value=30, 
+                max_value=1000, 
+                value=365,
                 help="Number of future time steps to predict"
             )
         
-        # Forecast button
+        # Forecast generation
         if st.button('Generate Forecast'):
             try:
-                # Initialize forecaster
+                # Initialize and run forecaster
                 forecaster = TimeSeriesForecaster()
                 
                 # Preprocess data
@@ -195,22 +212,17 @@ def main():
                 # Perform forecasting
                 forecasts = forecaster.forecast(context, prediction_length)
                 
-                # Visualization card
-                st.markdown("<div class='card'>", unsafe_allow_html=True)
-                st.subheader('Forecast Visualization')
-                plt = forecaster.visualize_forecast(context, forecasts)
+                # Visualization
+                plt = forecaster.visualize_forecast(forecasts, forecaster.original_series)
                 st.pyplot(plt)
-                st.markdown("</div>", unsafe_allow_html=True)
-                
-                # Forecast details card
-                st.markdown("<div class='card'>", unsafe_allow_html=True)
-                st.subheader('Forecast Details')
                 
+                # Forecast details
                 forecast_np = forecasts[0].numpy()
                 forecast_mean = forecast_np.mean(axis=0)
                 forecast_lower = np.percentile(forecast_np, 10, axis=0)
                 forecast_upper = np.percentile(forecast_np, 90, axis=0)
                 
+                st.subheader('Forecast Insights')
                 prediction_df = pd.DataFrame({
                     'Mean Forecast': forecast_mean,
                     'Lower Bound (10%)': forecast_lower,
@@ -218,12 +230,9 @@ def main():
                 })
                 
                 st.dataframe(prediction_df)
-                st.markdown("</div>", unsafe_allow_html=True)
                 
             except Exception as e:
-                st.error(f"An error occurred: {str(e)}")
-        
-        st.markdown("</div>", unsafe_allow_html=True)
+                st.error(f"Forecast generation error: {str(e)}")
 
 if __name__ == '__main__':
     main()