app.py

import streamlit as st
import time
import pandas as pd
import plotly.express as px
import plotly.graph_objects as go
import matplotlib.pyplot as plt
import numpy as np
import lightgbm as lgb
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity
from sklearn.metrics import mean_absolute_error, mean_squared_error
from joblib import dump, load
from utils import recomienda_tfid

# Page configuration
st.set_page_config(page_title="DeepInsightz", page_icon=":bar_chart:", layout="wide")

# Custom CSS for dynamic theme styling
# Streamlit detects light and dark mode automatically via the user's settings in Hugging Face Spaces
if st.get_option("theme.base") == "dark":
    background_color = "#282828"
    text_color = "white"
    metric_box_color = "#4f4f4f"
    sidebar_color = "#282828"
    plot_bgcolor = "rgba(0, 0, 0, 0)"
    primary_color = '#00FF00'  # for positive delta
    negative_color = '#FF0000'  # for negative delta
else:
    background_color = "#f4f4f4"
    text_color = "#black"
    metric_box_color = "#dee2e8"
    sidebar_color = "#dee2e8"
    plot_bgcolor = "#f4f4f4"
    primary_color = '#228B22'  # for positive delta in light mode
    negative_color = '#8B0000'  # for negative delta in light mode

st.markdown(f"""
    <style>
    body {{
        background-color: {background_color};
        color: {text_color};
    }}
    [data-testid="stMetric"] {{
        background-color: {metric_box_color};
        border-radius: 10px;
        text-align: center;
        padding: 15px 0;
        margin-bottom: 20px;
    }}
    [data-testid="stMetricLabel"] {{
        display: flex;
        justify-content: center;
        align-items: center;
        color: {text_color};
    }}
    [data-testid="stSidebar"] {{
        background-color: {sidebar_color};
    }}
    </style>
""", unsafe_allow_html=True)


# Load CSV files at the top
df = pd.read_csv("df_clean.csv")
nombres_proveedores = pd.read_csv("nombres_proveedores.csv", sep=';')
euros_proveedor = pd.read_csv("euros_proveedor.csv", sep=',')
ventas_clientes = pd.read_csv("ventas_clientes.csv", sep=',')
customer_clusters = pd.read_csv('predicts/customer_clusters.csv')  # Load the customer clusters here
df_agg_2024 = pd.read_csv('predicts/df_agg_2024.csv')
pca_data_5 = pd.read_csv('pca_data.csv')
historical_data = pd.read_csv('historical_data.csv')

with st.sidebar:
    st.sidebar.title("DeepInsightz")
    page = st.sidebar.selectbox("Select the tool you want to use", ["Summary", "Customer Analysis", "Articles Recommendations"])

    # If the user selects "Customer Analysis," show filter options in the sidebar
    if page == "Customer Analysis":
        st.sidebar.title("Filter Options")
        show_all = st.sidebar.checkbox('Show All Manufacturers', value=True)

        # If not showing all, allow filtering by manufacturer code
        if not show_all:
            selected_manufacturer = st.sidebar.selectbox(
                'Select Manufacturer Code', 
                historical_data['marca_id_encoded'].unique()
            )

# Generamos la columna total_sales
ventas_clientes['total_sales'] = ventas_clientes[['VENTA_2021', 'VENTA_2022', 'VENTA_2023']].sum(axis=1)
# Ordenar los clientes de mayor a menor según sus ventas totales
ventas_top_100 = ventas_clientes.sort_values(by='total_sales', ascending=False).head(100)


# Ensure customer codes are strings
df['CLIENTE'] = df['CLIENTE'].astype(str)
nombres_proveedores['codigo'] = nombres_proveedores['codigo'].astype(str)
euros_proveedor['CLIENTE'] = euros_proveedor['CLIENTE'].astype(str)
customer_clusters['cliente_id'] = customer_clusters['cliente_id'].astype(str)  # Ensure customer IDs are strings
fieles_df = pd.read_csv("clientes_relevantes.csv")
cestas = pd.read_csv("cestas.csv")
productos = pd.read_csv("productos.csv")
df_agg_2024['cliente_id'] = df_agg_2024['cliente_id'].astype(str)
marca_id_mapping = load('marca_id_mapping.joblib')

# Convert all columns except 'CLIENTE' to float in euros_proveedor
for col in euros_proveedor.columns:
    if col != 'CLIENTE':
        euros_proveedor[col] = pd.to_numeric(euros_proveedor[col], errors='coerce')

# Check for NaN values after conversion
if euros_proveedor.isna().any().any():
    st.warning("Some values in euros_proveedor couldn't be converted to numbers. Please review the input data.")

# Ignore the last two columns of df
df = df.iloc[:, :-2]

# Function to get supplier name
def get_supplier_name(code):
    code = str(code)  # Ensure code is a string
    name = nombres_proveedores[nombres_proveedores['codigo'] == code]['nombre'].values
    return name[0] if len(name) > 0 else code

def get_supplier_name_encoded(encoded_code):
    try:
        # Ensure the encoded code is an integer
        encoded_code = int(encoded_code)
        print(f"Encoded Code: {encoded_code}")

        # Use the label encoder to map the encoded code back to the original manufacturer code
        if encoded_code < len(marca_id_mapping.classes_):
            real_code = marca_id_mapping.inverse_transform([encoded_code])[0]
            print(f"Real Manufacturer Code: {real_code}")
        else:
            print(f"Encoded code not found in the label encoder: {encoded_code}")
            return f"Unknown code: {encoded_code}"  # Handle case where encoded code is not found

        # Now, use the real_code to find the manufacturer name in nombres_proveedores
        name = nombres_proveedores[nombres_proveedores['codigo'] == str(real_code)]['nombre'].values
        print(f"Manufacturer Name Found: {name}")  # Check what name is returned

        # Return the manufacturer name if found, otherwise return the real_code
        return name[0] if len(name) > 0 else real_code

    except Exception as e:
        print(f"Error encountered: {e}")
        return f"Error for code: {encoded_code}"

# Custom Donut Chart with Plotly for Inbound/Outbound Percentage
def create_donut_chart(values, labels, color_scheme, title):
    fig = px.pie(
        values=values, 
        names=labels, 
        hole=0.7,
        color_discrete_sequence=color_scheme
    )
    fig.update_traces(textinfo='percent+label', hoverinfo='label+percent', textposition='inside', showlegend=False)
    fig.update_layout(
        annotations=[dict(text=f"{int(values[1])}%", x=0.5, y=0.5, font_size=40, showarrow=False)],
        title=title,
        height=300,
        margin=dict(t=30, b=10, l=10, r=10),
        paper_bgcolor=plot_bgcolor,  # Use theme-dependent background color
        plot_bgcolor=plot_bgcolor
    )
    return fig

# Donut chart with color scheme based on theme
if st.get_option("theme.base") == "dark":
    donut_color_scheme = ['#155F7A', '#29b5e8']  # Dark mode colors
else:
    donut_color_scheme = ['#007BFF', '#66b5ff']  # Light mode colors

# Function to create radar chart with square root transformation
def radar_chart(categories, values, amounts, title):
    N = len(categories)
    angles = [n / float(N) * 2 * np.pi for n in range(N)]
    angles += angles[:1]
    
    fig, ax = plt.subplots(figsize=(12, 12), subplot_kw=dict(projection='polar'))
    
    # Apply square root transformation
    sqrt_values = np.sqrt(values)
    sqrt_amounts = np.sqrt(amounts)
    
    max_sqrt_value = max(sqrt_values)
    normalized_values = [v / max_sqrt_value for v in sqrt_values]
    
    # Adjust scaling for spend values
    max_sqrt_amount = max(sqrt_amounts)
    scaling_factor = 0.7  # Adjust this value to control how much the spend values are scaled up
    normalized_amounts = [min((a / max_sqrt_amount) * scaling_factor, 1.0) for a in sqrt_amounts]
    
    normalized_values += normalized_values[:1]
    ax.plot(angles, normalized_values, 'o-', linewidth=2, color='#FF69B4', label='% Units (sqrt)')
    ax.fill(angles, normalized_values, alpha=0.25, color='#FF69B4')
    
    normalized_amounts += normalized_amounts[:1]
    ax.plot(angles, normalized_amounts, 'o-', linewidth=2, color='#4B0082', label='% Spend (sqrt)')
    ax.fill(angles, normalized_amounts, alpha=0.25, color='#4B0082')
    
    ax.set_xticks(angles[:-1])
    ax.set_xticklabels(categories, size=8, wrap=True)
    ax.set_ylim(0, 1)
    
    circles = np.linspace(0, 1, 5)
    for circle in circles:
        ax.plot(angles, [circle]*len(angles), '--', color='gray', alpha=0.3, linewidth=0.5)
    
    ax.set_yticklabels([])
    ax.spines['polar'].set_visible(False)
    
    plt.title(title, size=16, y=1.1)
    plt.legend(loc='upper right', bbox_to_anchor=(1.3, 1.1))
    
    return fig



if page == "Summary":
    # st.title("Welcome to DeepInsightz")
    # st.markdown("""
    #     ### Data-driven Customer Clustering
    #     We analyzed thousands of customers and suppliers to help businesses make smarter sales decisions.
    # """)

    # Create layout with three columns
    col1, col2, col3 = st.columns((1.5, 4, 2.5), gap='medium')

    # Left Column (Red): Metrics and Donut Charts
    with col1:
        st.markdown('#### General Information')
        st.metric(label="Range of Dates", value="2021-2023")
        st.metric(label="Customers Analysed", value="3.000")
        st.metric(label="Unique Products Sold", value="10.702")
        st.metric(label="Total Sales Instances", value="764.396")
        


    # Middle Column (White): 3D Cluster Model and Bar Chart
    with col2:
        st.markdown('#### 3D Customer Clusters')

        # Create 3D PCA plot using actual data from pca_data_5
        fig_cluster = px.scatter_3d(
            pca_data_5, 
            x='PC1', 
            y='PC2', 
            z='PC3', 
            color='cluster_id', 
            hover_name='CustomerID',
        )
        fig_cluster.update_layout(
            scene=dict(aspectratio=dict(x=1, y=1, z=0.8)),  # Adjusted aspect ratio for better balance
            margin=dict(t=10, b=10, l=10, r=10),  # Tighten margins further
            height=600,  # Slightly increased height for better visibility
        )
        st.plotly_chart(fig_cluster, use_container_width=True)
    
    # Right Column (Blue): Key Metrics Overview and Data Preparation Summary
    with col3:
        # Mostrar la tabla con los 100 mejores clientes
        st.markdown('#### Top 100 Clients by Total Sales')

        # Configurar columnas para mostrar los clientes y las ventas totales
        st.dataframe(ventas_top_100[['codigo_cliente', 'total_sales']],
                    column_order=("codigo_cliente", "total_sales"),
                    hide_index=True,
                    width=450,  # Ajustar el ancho de la tabla
                    height=600,  # Ajustar la altura de la tabla
                    column_config={
                        "codigo_cliente": st.column_config.TextColumn(
                            "Client Code",
                        ),
                        "total_sales": st.column_config.ProgressColumn(
                            "Total Sales (€)",
                            format="%d",
                            min_value=0,
                            max_value=ventas_top_100['total_sales'].max()
                        )}
                    )
# Customer Analysis Page
elif page == "Customer Analysis":
    st.markdown("""
    <h2 style='text-align: center; font-size: 2.5rem;'>Customer Analysis</h2>
    <p style='text-align: center; font-size: 1.2rem; color: gray;'> 
    Enter the customer code to explore detailed customer insights, 
    including past sales, predictions for the current year, and manufacturer-specific information.
    </p>
    """, unsafe_allow_html=True)

    # Combine text input and dropdown into a single searchable selectbox
    customer_code = st.selectbox(
        "Search and Select Customer Code",
        df['CLIENTE'].unique(),  # All customer codes

        format_func=lambda x: str(x),  # Ensures the values are displayed as strings
        help="Start typing to search for a specific customer code"
    )

    if st.button("Calcular"):
        if customer_code:
            with st.spinner("We are identifying the customer's cluster..."):
                # Find Customer's Cluster
                customer_match = customer_clusters[customer_clusters['cliente_id'] == customer_code]
                time.sleep(1)
                
                if not customer_match.empty:
                    cluster = customer_match['cluster_id'].values[0]
                
            with st.spinner(f"Selecting predictive model..."):
                # Load the Corresponding Model
                model_path = f'models/modelo_cluster_{cluster}.txt'
                gbm = lgb.Booster(model_file=model_path)

            with st.spinner("Getting the data ready..."):
                # Load predict data for that cluster
                predict_data = pd.read_csv(f'predicts/predict_cluster_{cluster}.csv')
                
                # Convert cliente_id to string
                predict_data['cliente_id'] = predict_data['cliente_id'].astype(str)

            with st.spinner("Filtering data..."):
                # Filter for the specific customer
                customer_code_str = str(customer_code)
                customer_data = predict_data[predict_data['cliente_id'] == customer_code_str]

            with st.spinner("Generating sales predictions..."):
                if not customer_data.empty:
                    # Define features consistently with the training process
                    lag_features = [f'precio_total_lag_{lag}' for lag in range(1, 25)]
                    features = lag_features + ['mes', 'marca_id_encoded', 'año', 'cluster_id']
    
                    # Prepare data for prediction
                    X_predict = customer_data[features]

                    # Convert categorical features to 'category' dtype
                    categorical_features = ['mes', 'marca_id_encoded', 'cluster_id']
                    for feature in categorical_features:
                        X_predict[feature] = X_predict[feature].astype('category')
                    
                    # Make Prediction for the selected customer
                    y_pred = gbm.predict(X_predict, num_iteration=gbm.best_iteration)

                    # Reassemble the results
                    results = customer_data[['cliente_id', 'marca_id_encoded', 'fecha_mes']].copy()
                    results['ventas_predichas'] = y_pred

                    # Load actual data from df_agg_2024
                    actual_sales = df_agg_2024[df_agg_2024['cliente_id'] == customer_code_str]
                    
                    if not actual_sales.empty:
                        # Merge predictions with actual sales
                        results = results.merge(actual_sales[['cliente_id', 'marca_id_encoded', 'fecha_mes', 'precio_total']], 
                                                on=['cliente_id', 'marca_id_encoded', 'fecha_mes'], 
                                                how='left')
                        results.rename(columns={'precio_total': 'ventas_reales'}, inplace=True)
                    else:
                        # If no actual sales data for 2024, fill 'ventas_reales' with 0
                        results['ventas_reales'] = 0

                    # Ensure any missing sales data is filled with 0
                    results['ventas_reales'].fillna(0, inplace=True)

                    # Define the cutoff date for the last 12 months
                    fecha_inicio = pd.to_datetime("2023-01-01")
                    fecha_corte = pd.to_datetime("2024-09-01")

                    # Convertir fecha_mes a datetime en el DataFrame historical_data
                    historical_data['fecha_mes'] = pd.to_datetime(historical_data['fecha_mes'], errors='coerce')

                    # Ensure cliente_id is of type string and strip any leading/trailing whitespace
                    historical_data['cliente_id'] = historical_data['cliente_id'].astype(str).str.strip()
                    customer_code_str = str(customer_code).strip()  # Ensure the customer code is also properly formatted

                    filtered_historical_data = historical_data[historical_data['cliente_id'] == customer_code_str]


                    # Filtrar los datos históricos por cliente y por el rango de fechas (2023)
                    fecha_inicio_2023 = pd.to_datetime("2023-01-01")
                    fecha_fin_2023 = pd.to_datetime("2023-12-31")

                    datos_historicos = historical_data[
                        (historical_data['cliente_id'] == customer_code_str) &
                        (historical_data['fecha_mes'] >= fecha_inicio_2023) &
                        (historical_data['fecha_mes'] <= fecha_fin_2023)
                    ].groupby('fecha_mes')['precio_total'].sum().reset_index()

                    # Renombrar la columna 'precio_total' a 'ventas_historicas' si no está vacía
                    if not datos_historicos.empty:
                        datos_historicos.rename(columns={'precio_total': 'ventas_historicas'}, inplace=True)
                    else:
                        # Si los datos históricos están vacíos, generar fechas de 2023 con ventas_historicas = 0
                        fechas_2023 = pd.date_range(start='2023-01-01', end='2023-12-31', freq='M')
                        datos_historicos = pd.DataFrame({'fecha_mes': fechas_2023, 'ventas_historicas': [0] * len(fechas_2023)})

                    # Filtrar los datos de predicciones y ventas reales para 2024
                    datos_cliente_total = results.groupby('fecha_mes').agg({
                        'ventas_reales': 'sum',
                        'ventas_predichas': 'sum'
                    }).reset_index()

                    # Asegurarnos de que fecha_mes en datos_cliente_total es datetime
                    datos_cliente_total['fecha_mes'] = pd.to_datetime(datos_cliente_total['fecha_mes'], errors='coerce')

                    # Generar un rango de fechas para 2024 si no hay predicciones
                    fechas_2024 = pd.date_range(start='2024-01-01', end='2024-12-31', freq='M')
                    fechas_df_2024 = pd.DataFrame({'fecha_mes': fechas_2024})

                    # Asegurarnos de que fecha_mes en fechas_df_2024 es datetime
                    fechas_df_2024['fecha_mes'] = pd.to_datetime(fechas_df_2024['fecha_mes'], errors='coerce')

                    # Combinar datos históricos con predicciones y ventas reales usando un merge
                    # Usamos how='outer' para asegurarnos de incluir todas las fechas de 2023 y 2024
                    datos_combinados = pd.merge(datos_historicos, datos_cliente_total, on='fecha_mes', how='outer').sort_values('fecha_mes')

                    # Rellenar los NaN: 0 en ventas_historicas donde faltan predicciones, y viceversa
                    datos_combinados['ventas_historicas'].fillna(0, inplace=True)
                    datos_combinados['ventas_predichas'].fillna(0, inplace=True)
                    datos_combinados['ventas_reales'].fillna(0, inplace=True)

                    # Crear la gráfica con Plotly
                    fig = go.Figure()

                    # Graficar ventas históricas
                    fig.add_trace(go.Scatter(
                        x=datos_combinados['fecha_mes'], 
                        y=datos_combinados['ventas_historicas'], 
                        mode='lines+markers', 
                        name='Ventas Históricas', 
                        line=dict(color='blue')
                    ))

                    # Graficar ventas predichas
                    fig.add_trace(go.Scatter(
                        x=datos_combinados['fecha_mes'], 
                        y=datos_combinados['ventas_predichas'], 
                        mode='lines+markers', 
                        name='Ventas Predichas', 
                        line=dict(color='orange')
                    ))

                    # Graficar ventas reales
                    fig.add_trace(go.Scatter(
                        x=datos_combinados['fecha_mes'], 
                        y=datos_combinados['ventas_reales'], 
                        mode='lines+markers', 
                        name='Ventas Reales', 
                        line=dict(color='green')
                    ))

                    # Personalizar el layout para enfocarse en 2023 y 2024
                    fig.update_layout(
                        title=f"Ventas Históricas, Predichas y Reales para Cliente {customer_code}",
                        xaxis_title="Fecha",
                        yaxis_title="Ventas (€)",
                        height=600,
                        xaxis_range=[fecha_inicio_2023, pd.to_datetime("2024-09-30")],  # Ajustar el rango del eje x a 2023-2024
                        legend_title="Tipo de Ventas",
                        hovermode="x unified"
                    )

                    # Mostrar la gráfica en Streamlit
                    st.plotly_chart(fig)


                    # Split space into two columns
                    col1, col2 = st.columns(2)

                    # Column 1: Radar chart for top manufacturers
                    with col1:
                        st.subheader("This looks great!")
                        st.info("Your customer did exceed predicted sales from the following brands:")

                        # Group results by manufacturer to calculate the total predicted and actual sales
                        grouped_results = results.groupby('marca_id_encoded').agg({
                            'ventas_reales': 'sum',
                            'ventas_predichas': 'sum'
                        }).reset_index()

                        # Identify manufacturers that exceeded predicted sales
                        overperforming_manufacturers = grouped_results[grouped_results['ventas_reales'] > grouped_results['ventas_predichas']].copy()

                        if not overperforming_manufacturers.empty:
                            # Calculate the extra amount (difference between actual and predicted sales)
                            overperforming_manufacturers['extra_amount'] = overperforming_manufacturers['ventas_reales'] - overperforming_manufacturers['ventas_predichas']

                            # Sort by the highest extra amount
                            overperforming_manufacturers = overperforming_manufacturers.sort_values(by='extra_amount', ascending=False)

                            # Limit to top 10 overperforming manufacturers
                            top_overperformers = overperforming_manufacturers.head(10)

                            # Display two cards per row
                            for i in range(0, len(top_overperformers), 2):
                                cols = st.columns(2)  # Create two columns for two cards in a row

                                for j, col in enumerate(cols):
                                    if i + j < len(top_overperformers):
                                        row = top_overperformers.iloc[i + j]
                                        manufacturer_name = get_supplier_name_encoded(row['marca_id_encoded'])
                                        predicted = row['ventas_predichas']
                                        actual = row['ventas_reales']
                                        extra = row['extra_amount']

                                        # Use st.metric for compact display in each column
                                        with col:
                                            st.metric(
                                                label=f"{manufacturer_name}",
                                                value=f"{actual:.2f}€",
                                                delta=f"Exceeded by {extra:.2f}€",
                                                delta_color="normal"
                                            )


                        # Radar chart logic remains the same
                        customer_df = df[df["CLIENTE"] == str(customer_code)]
                        all_manufacturers = customer_df.iloc[:, 1:].T
                        all_manufacturers.index = all_manufacturers.index.astype(str)

                        customer_euros = euros_proveedor[euros_proveedor["CLIENTE"] == str(customer_code)]
                        sales_data = customer_euros.iloc[:, 1:].T
                        sales_data.index = sales_data.index.astype(str)

                        sales_data_filtered = sales_data.drop(index='CLIENTE', errors='ignore')
                        sales_data_filtered = sales_data_filtered.apply(pd.to_numeric, errors='coerce')
                        all_manufacturers = all_manufacturers.apply(pd.to_numeric, errors='coerce')

                        top_units = all_manufacturers.sort_values(by=all_manufacturers.columns[0], ascending=False).head(10)
                        top_sales = sales_data_filtered.sort_values(by=sales_data_filtered.columns[0], ascending=False).head(10)
                        combined_top = pd.concat([top_units, top_sales]).index.unique()[:20]

                        combined_top = [m for m in combined_top if m in all_manufacturers.index and m in sales_data_filtered.index]

                        if combined_top:
                            combined_data = pd.DataFrame({
                                'units': all_manufacturers.loc[combined_top, all_manufacturers.columns[0]],
                                'sales': sales_data_filtered.loc[combined_top, sales_data_filtered.columns[0]]
                            }).fillna(0)

                            combined_data_sorted = combined_data.sort_values(by=['units', 'sales'], ascending=False)
                            non_zero_manufacturers = combined_data_sorted[combined_data_sorted['units'] > 0]
                            
                            if len(non_zero_manufacturers) < 3:
                                zero_manufacturers = combined_data_sorted[combined_data_sorted['units'] == 0].head(3 - len(non_zero_manufacturers))
                                manufacturers_to_show = pd.concat([non_zero_manufacturers, zero_manufacturers])
                            else:
                                manufacturers_to_show = non_zero_manufacturers

                            values = manufacturers_to_show['units'].tolist()
                            amounts = manufacturers_to_show['sales'].tolist()
                            manufacturers = [get_supplier_name(m) for m in manufacturers_to_show.index]

                            if manufacturers:
                                fig = radar_chart(manufacturers, values, amounts, f'Radar Chart for Top {len(manufacturers)} Manufacturers of Customer {customer_code}')
                                st.pyplot(fig)

                    # Column 2: Alerts and additional analysis
                    with col2:
                        st.subheader("You might need to check this!")
                        st.warning("Your customer was expected to buy more products from the following brands:")

                        # Group results by manufacturer to calculate the total predicted and actual sales
                        grouped_results = results.groupby('marca_id_encoded').agg({
                            'ventas_reales': 'sum',
                            'ventas_predichas': 'sum'
                        }).reset_index()

                        # Identify manufacturers that didn't meet predicted sales
                        underperforming_manufacturers = grouped_results[grouped_results['ventas_reales'] < grouped_results['ventas_predichas']].copy()

                        if not underperforming_manufacturers.empty:
                            # Calculate the missed amount
                            underperforming_manufacturers['missed_amount'] = underperforming_manufacturers['ventas_predichas'] - underperforming_manufacturers['ventas_reales']

                            # Sort by the highest missed amount
                            underperforming_manufacturers = underperforming_manufacturers.sort_values(by='missed_amount', ascending=False)

                            # Limit to top 10 missed amounts
                            top_misses = underperforming_manufacturers.head(10)

                            # Display two cards per row
                            for i in range(0, len(top_misses), 2):
                                cols = st.columns(2)  # Create two columns for two cards in a row

                                for j, col in enumerate(cols):
                                    if i + j < len(top_misses):
                                        row = top_misses.iloc[i + j]
                                        manufacturer_name = get_supplier_name_encoded(row['marca_id_encoded'])
                                        predicted = row['ventas_predichas']
                                        actual = row['ventas_reales']
                                        missed = row['missed_amount']

                                        # Use st.metric for compact display in each column
                                        with col:
                                            st.metric(
                                                label=f"{manufacturer_name}",
                                                value=f"{actual:.2f}€",
                                                delta=f"Missed by {missed:.2f}€",
                                                delta_color="inverse"
                                            )
                        else:
                            st.success("All manufacturers have met or exceeded predicted sales.")

                        # Gráfico adicional: Comparar las ventas predichas y reales para los principales fabricantes
                        st.markdown("### Predicted vs Actual Sales for Top Manufacturers")
                        top_manufacturers = results.groupby('marca_id_encoded').agg({'ventas_reales': 'sum', 'ventas_predichas': 'sum'}).sort_values(by='ventas_reales', ascending=False).head(10)

                        fig_comparison = go.Figure()
                        fig_comparison.add_trace(go.Bar(x=top_manufacturers.index, y=top_manufacturers['ventas_reales'], name="Actual Sales", marker_color='blue'))
                        fig_comparison.add_trace(go.Bar(x=top_manufacturers.index, y=top_manufacturers['ventas_predichas'], name="Predicted Sales", marker_color='orange'))

                        fig_comparison.update_layout(
                            title="Actual vs Predicted Sales by Top Manufacturers",
                            xaxis_title="Manufacturer",
                            yaxis_title="Sales (€)",
                            barmode='group',
                            height=400,
                            hovermode="x unified"
                        )

                        st.plotly_chart(fig_comparison, use_container_width=True)

                        # Gráfico de ventas anuales
                        ventas_clientes['codigo_cliente'] = ventas_clientes['codigo_cliente'].astype(str).str.strip()

                        sales_columns = ['VENTA_2021', 'VENTA_2022', 'VENTA_2023']
                        if all(col in ventas_clientes.columns for col in sales_columns):
                            customer_sales_data = ventas_clientes[ventas_clientes['codigo_cliente'] == customer_code]

                            if not customer_sales_data.empty:
                                customer_sales = customer_sales_data[sales_columns].values[0]
                                years = ['2021', '2022', '2023']

                                # Convert 'fecha_mes' to datetime format if it's not already
                                if not pd.api.types.is_datetime64_any_dtype(results['fecha_mes']):
                                    results['fecha_mes'] = pd.to_datetime(results['fecha_mes'], errors='coerce')

                                # Add the 2024 actual and predicted data
                                if 'ventas_predichas' in results.columns and 'ventas_reales' in results.columns:
                                    actual_sales_2024 = results[results['fecha_mes'].dt.year == 2024]['ventas_reales'].sum()
                                    predicted_sales_2024 = results[results['fecha_mes'].dt.year == 2024]['ventas_predichas'].sum()

                                    # Assuming only 9 months of actual data are available, annualize the sales
                                    months_available = 9
                                    actual_sales_2024_annual = (actual_sales_2024 / months_available) * 12

                                    # Prepare data for the bar chart
                                    sales_values = list(customer_sales) + [actual_sales_2024_annual]
                                    predicted_values = list(customer_sales) + [predicted_sales_2024]

                                    years.append('2024')

                                    # Create the bar chart for historical and 2024 data
                                    fig_sales_bar = go.Figure()
                                    fig_sales_bar.add_trace(go.Bar(
                                        x=years[:3],  
                                        y=sales_values[:3],
                                        name="Historical Sales",
                                        marker_color='blue'
                                    ))

                                    fig_sales_bar.add_trace(go.Bar(
                                        x=[years[3]],  
                                        y=[sales_values[3]],
                                        name="2024 Actual Sales (Annualized)",
                                        marker_color='green'
                                    ))

                                    fig_sales_bar.add_trace(go.Bar(
                                        x=[years[3]],  
                                        y=[predicted_values[3]],
                                        name="2024 Predicted Sales",
                                        marker_color='orange'
                                    ))

                                    # Customize layout
                                    fig_sales_bar.update_layout(
                                        title=f"Sales Over the Years for Customer {customer_code}",
                                        xaxis_title="Year",
                                        yaxis_title="Sales (€)",
                                        barmode='group',
                                        height=600,
                                        legend_title_text="Sales Type",
                                        hovermode="x unified"
                                    )

                                    # Display the chart
                                    st.plotly_chart(fig_sales_bar, use_container_width=True)

                                else:
                                    st.warning(f"No predicted or actual data found for customer {customer_code} for 2024.")



# elif page == "Customer Analysis":
#     st.markdown("""
#     <h2 style='text-align: center; font-size: 2.5rem;'>Customer Analysis</h2>
#     <p style='text-align: center; font-size: 1.2rem; color: gray;'> 
#     Enter the customer code to explore detailed customer insights, 
#     including past sales, predictions for the current year, and manufacturer-specific information.
#     </p>
#     """, unsafe_allow_html=True)

#     # Combine text input and dropdown into a single searchable selectbox
#     customer_code = st.selectbox(
#         "Search and Select Customer Code",
#         df['CLIENTE'].unique(),  # All customer codes
#         format_func=lambda x: str(x),  # Ensures the values are displayed as strings
#         help="Start typing to search for a specific customer code"
#     )

#     if st.button("Calcular"):
#         if customer_code:
#             with st.spinner("We are identifying the customer's cluster..."):
#                 # Find Customer's Cluster
#                 customer_match = customer_clusters[customer_clusters['cliente_id'] == customer_code]
#                 time.sleep(1)
                

#                 if not customer_match.empty:
#                     cluster = customer_match['cluster_id'].values[0]
                
#             with st.spinner(f"Selecting predictive model..."):
#                 # Load the Corresponding Model
#                 model_path = f'models/modelo_cluster_{cluster}.txt'
#                 gbm = lgb.Booster(model_file=model_path)

#             with st.spinner("Getting the data ready..."):
#                 # Load predict data for that cluster
#                 predict_data = pd.read_csv(f'predicts/predict_cluster_{cluster}.csv')
                
#                 # Convert cliente_id to string
#                 predict_data['cliente_id'] = predict_data['cliente_id'].astype(str)

#             with st.spinner("Filtering data..."):                

#                 # Filter for the specific customer
#                 customer_code_str = str(customer_code)
#                 customer_data = predict_data[predict_data['cliente_id'] == customer_code_str]

#             with st.spinner("Generating sales predictions..."):       

#                 if not customer_data.empty:
#                     # Define features consistently with the training process
#                     lag_features = [f'precio_total_lag_{lag}' for lag in range(1, 25)]
#                     features = lag_features + ['mes', 'marca_id_encoded', 'año', 'cluster_id']
    
#                     # Prepare data for prediction
#                     X_predict = customer_data[features]

#                     # Convert categorical features to 'category' dtype
#                     categorical_features = ['mes', 'marca_id_encoded', 'cluster_id']
#                     for feature in categorical_features:
#                         X_predict[feature] = X_predict[feature].astype('category')
                    
#                     # Make Prediction for the selected customer
#                     y_pred = gbm.predict(X_predict, num_iteration=gbm.best_iteration)

#                     # Reassemble the results
#                     results = customer_data[['cliente_id', 'marca_id_encoded', 'fecha_mes']].copy()
#                     results['ventas_predichas'] = y_pred

#                     # Load actual data
#                     actual_sales = df_agg_2024[df_agg_2024['cliente_id'] == customer_code_str]
                    
#                     if not actual_sales.empty:
#                         results = results.merge(actual_sales[['cliente_id', 'marca_id_encoded', 'fecha_mes', 'precio_total']], 
#                                                 on=['cliente_id', 'marca_id_encoded', 'fecha_mes'], 
#                                                 how='left')
#                         results.rename(columns={'precio_total': 'ventas_reales'}, inplace=True)
#                         results['ventas_reales'].fillna(0, inplace=True)
#                         # st.write("### Final Results DataFrame:")
#                         # st.write(results.head())
#                         # st.write(f"Shape: {results.shape}")
                        
#                         # Calculate metrics only for non-null actual sales
#                         valid_results = results.dropna(subset=['ventas_reales'])
#                         non_zero_actuals = valid_results[valid_results['ventas_reales'] != 0]
#                         if not valid_results.empty:
#                             mae = mean_absolute_error(valid_results['ventas_reales'], valid_results['ventas_predichas'])
#                             mape = np.mean(np.abs((non_zero_actuals['ventas_reales'] - non_zero_actuals['ventas_predichas']) / non_zero_actuals['ventas_reales'])) * 100
#                             rmse = np.sqrt(mean_squared_error(valid_results['ventas_reales'], valid_results['ventas_predichas']))

#                             # st.write(f"Actual total sales for Customer {customer_code}: {valid_results['ventas_reales'].sum():.2f}")
#                             # st.write(f"MAE: {mae:.2f}€")
#                             # st.write(f"MAPE: {mape:.2f}%")
#                             # st.write(f"RMSE: {rmse:.2f}")

#                     #     # Analysis of results
#                     #     threshold_good = 100  # You may want to adjust this threshold
#                     #     if mae < threshold_good:
#                     #         st.success(f"Customer {customer_code} is performing well based on the predictions.")
#                     #     else:
#                     #         st.warning(f"Customer {customer_code} is not performing well based on the predictions.")
#                     # else:
#                     #     st.warning(f"No actual sales data found for customer {customer_code} in df_agg_2024.")

#                     # st.write("### Debug Information for Radar Chart:")
#                     # st.write(f"Shape of customer_data: {customer_data.shape}")
#                     # st.write(f"Shape of euros_proveedor: {euros_proveedor.shape}")

#                     # Get percentage of units sold for each manufacturer
#                     customer_df = df[df["CLIENTE"] == str(customer_code)]  # Get the customer data
#                     all_manufacturers = customer_df.iloc[:, 1:].T  # Exclude CLIENTE column (manufacturers are in columns)
#                     all_manufacturers.index = all_manufacturers.index.astype(str)

#                     # Get total sales for each manufacturer from euros_proveedor
#                     customer_euros = euros_proveedor[euros_proveedor["CLIENTE"] == str(customer_code)]
#                     sales_data = customer_euros.iloc[:, 1:].T  # Exclude CLIENTE column
#                     sales_data.index = sales_data.index.astype(str)

#                     # Remove the 'CLIENTE' row from sales_data to avoid issues with mixed types
#                     sales_data_filtered = sales_data.drop(index='CLIENTE', errors='ignore')

#                     # Ensure all values are numeric
#                     sales_data_filtered = sales_data_filtered.apply(pd.to_numeric, errors='coerce')
#                     all_manufacturers = all_manufacturers.apply(pd.to_numeric, errors='coerce')

#                     # Sort manufacturers by percentage of units and get top 10
#                     top_units = all_manufacturers.sort_values(by=all_manufacturers.columns[0], ascending=False).head(10)

#                     # Sort manufacturers by total sales and get top 10
#                     top_sales = sales_data_filtered.sort_values(by=sales_data_filtered.columns[0], ascending=False).head(10)

#                     # Combine top manufacturers from both lists and get up to 20 unique manufacturers
#                     combined_top = pd.concat([top_units, top_sales]).index.unique()[:20]

#                     # Filter out manufacturers that are not present in both datasets
#                     combined_top = [m for m in combined_top if m in all_manufacturers.index and m in sales_data_filtered.index]

#                     # st.write(f"Number of combined top manufacturers: {len(combined_top)}")

#                     if combined_top:
#                         # Create a DataFrame with combined data for these top manufacturers
#                         combined_data = pd.DataFrame({
#                             'units': all_manufacturers.loc[combined_top, all_manufacturers.columns[0]],
#                             'sales': sales_data_filtered.loc[combined_top, sales_data_filtered.columns[0]]
#                         }).fillna(0)

#                         # Sort by units, then by sales
#                         combined_data_sorted = combined_data.sort_values(by=['units', 'sales'], ascending=False)

#                         # Filter out manufacturers with 0 units
#                         non_zero_manufacturers = combined_data_sorted[combined_data_sorted['units'] > 0]

#                         # If we have less than 3 non-zero manufacturers, add some zero-value ones
#                         if len(non_zero_manufacturers) < 3:
#                             zero_manufacturers = combined_data_sorted[combined_data_sorted['units'] == 0].head(3 - len(non_zero_manufacturers))
#                             manufacturers_to_show = pd.concat([non_zero_manufacturers, zero_manufacturers])
#                         else:
#                             manufacturers_to_show = non_zero_manufacturers

#                         values = manufacturers_to_show['units'].tolist()
#                         amounts = manufacturers_to_show['sales'].tolist()
#                         manufacturers = [get_supplier_name(m) for m in manufacturers_to_show.index]

#                         # st.write(f"### Results for top {len(manufacturers)} manufacturers:")
#                         # for manufacturer, value, amount in zip(manufacturers, values, amounts):
#                         #     (f"{manufacturer} = {value:.2f}% of units, €{amount:.2f} total sales")

#                         if manufacturers:  # Only create the chart if we have data
#                             fig = radar_chart(manufacturers, values, amounts, f'Radar Chart for Top {len(manufacturers)} Manufacturers of Customer {customer_code}')
#                             st.pyplot(fig)
#                         else:
#                             st.warning("No data available to create the radar chart.")

#                     else:
#                         st.warning("No combined top manufacturers found.")

#                     # Ensure codigo_cliente in ventas_clientes is a string
#                     ventas_clientes['codigo_cliente'] = ventas_clientes['codigo_cliente'].astype(str).str.strip()

#                     # Ensure customer_code is a string and strip any spaces
#                     customer_code = str(customer_code).strip()

#                     # if customer_code in ventas_clientes['codigo_cliente'].unique():
#                     #     (f"Customer {customer_code} found in ventas_clientes")  
#                     # else:
#                     #     (f"Customer {customer_code} not found in ventas_clientes")

#                     # Customer sales 2021-2024 (if data exists)
#                     sales_columns = ['VENTA_2021', 'VENTA_2022', 'VENTA_2023']
#                     if all(col in ventas_clientes.columns for col in sales_columns):
#                         customer_sales_data = ventas_clientes[ventas_clientes['codigo_cliente'] == customer_code]

#                         if not customer_sales_data.empty:
#                             customer_sales = customer_sales_data[sales_columns].values[0]
#                             years = ['2021', '2022', '2023']

#                             # Add the 2024 actual and predicted data
#                             if 'ventas_predichas' in results.columns and 'ventas_reales' in results.columns:
#                                 # Get the actual and predicted sales for 2024
#                                 actual_sales_2024 = results[results['fecha_mes'].str.startswith('2024')]['ventas_reales'].sum()
#                                 predicted_sales_2024 = results[results['fecha_mes'].str.startswith('2024')]['ventas_predichas'].sum()

#                                 # Estimate full-year predicted sales (assuming predictions available until September)
#                                 months_available = 9  # Data available until September
#                                 actual_sales_2024_annual = (actual_sales_2024 / months_available) * 12

#                                 # Add 2024 actual and predicted sales
#                                 sales_values = list(customer_sales) + [actual_sales_2024_annual]  # Actual sales
#                                 predicted_values = list(customer_sales) + [predicted_sales_2024]  # Predicted sales

#                                 # Add 2024 to the years list
#                                 years.append('2024')

#                                 fig_sales_bar = go.Figure()
#                                 # Add trace for historical sales (2021-2023)
#                                 fig_sales_bar.add_trace(go.Bar(
#                                     x=years[:3],  # 2021, 2022, 2023
#                                     y=sales_values[:3],
#                                     name="Historical Sales",
#                                     marker_color='blue'
#                                 ))

#                                 # Add trace for 2024 actual sales
#                                 fig_sales_bar.add_trace(go.Bar(
#                                     x=[years[3]],  # 2024
#                                     y=[sales_values[3]],
#                                     name="2024 Actual Sales (Annualized)",
#                                     marker_color='green'
#                                 ))

#                                 # Add trace for 2024 predicted sales
#                                 fig_sales_bar.add_trace(go.Bar(
#                                     x=[years[3]],  # 2024
#                                     y=[predicted_values[3]],
#                                     name="2024 Predicted Sales",
#                                     marker_color='orange'
#                                 ))

#                                 # Update layout
#                                 fig_sales_bar.update_layout(
#                                     title=f"Sales Over the Years for Customer {customer_code}",
#                                     xaxis_title="Year",
#                                     yaxis_title="Sales (€)",
#                                     barmode='group',
#                                     height=600,
#                                     legend_title_text="Sales Type",
#                                     hovermode="x unified"
#                                 )

#                                 # Show the interactive bar chart in Streamlit
#                                 st.plotly_chart(fig_sales_bar, use_container_width=True)

#                             else:
#                                 st.warning(f"No predicted or actual data found for customer {customer_code} for 2024.")

#                         else:
#                             st.warning(f"No historical sales data found for customer {customer_code}")

#                     else:
#                         st.warning("Sales data for 2021-2023 not available in the dataset.")


# Customer Recommendations Page
elif page == "Articles Recommendations":
    st.title("Articles Recommendations")

    st.markdown("""
        Get tailored recommendations for your customers based on their basket.
    """)

    st.write("Select items and assign quantities for the basket:")

    # Mostrar lista de artículos disponibles
    available_articles = productos['ARTICULO'].unique()
    selected_articles = st.multiselect("Select Articles", available_articles)

    # Crear inputs para ingresar las cantidades de cada artículo seleccionado
    quantities = {}
    for article in selected_articles:
        quantities[article] = st.number_input(f"Quantity for {article}", min_value=0, step=1)

    if st.button("Calcular"):  # Añadimos el botón "Calcular"
                # Crear una lista de artículos basada en la selección
        new_basket = [f"{article} x{quantities[article]}" for article in selected_articles if quantities[article] > 0]

        if new_basket:
            # Procesar la lista para recomendar
            recommendations_df = recomienda_tfid(new_basket)

            if not recommendations_df.empty:
                st.write("### Recommendations based on the current basket:")
                st.dataframe(recommendations_df)
            else:
                st.warning("No recommendations found for the provided basket.")
        else:
            st.warning("Please select at least one article and set its quantity.")

# # Customer Recommendations Page
# elif page == "Articles Recommendations":
#     st.title("Articles Recommendations")

#     st.markdown("""
#         Get tailored recommendations for your customers based on their basket.
#     """)

#     # Campo input para cliente
#     partial_code = st.text_input("Enter part of Customer Code for Recommendations (or leave empty to see all)")
#     if partial_code:
#         filtered_customers = df[df['CLIENTE'].str.contains(partial_code)]
#     else:
#         filtered_customers = df
#     customer_list = filtered_customers['CLIENTE'].unique()
#     customer_code = st.selectbox("Select Customer Code for Recommendations", [""] + list(customer_list))

#     # Definición de la función recomienda
#     def recomienda(new_basket):
#         # Calcular la matriz TF-IDF
#         tfidf = TfidfVectorizer()
#         tfidf_matrix = tfidf.fit_transform(cestas['Cestas'])

#         # Convertir la nueva cesta en formato TF-IDF
#         new_basket_str = ' '.join(new_basket)
#         new_basket_tfidf = tfidf.transform([new_basket_str])

#         # Comparar la nueva cesta con las anteriores
#         similarities = cosine_similarity(new_basket_tfidf, tfidf_matrix)

#         # Obtener los índices de las cestas más similares
#         similar_indices = similarities.argsort()[0][-3:]  # Las 3 más similares

#         # Crear un diccionario para contar las recomendaciones
#         recommendations_count = {}
#         total_similarity = 0

#         # Recomendar productos de cestas similares
#         for idx in similar_indices:
#             sim_score = similarities[0][idx]
#             total_similarity += sim_score
#             products = cestas.iloc[idx]['Cestas'].split()

#             for product in products:
#                 if product.strip() not in new_basket:  # Evitar recomendar lo que ya está en la cesta
#                     if product.strip() in recommendations_count:
#                         recommendations_count[product.strip()] += sim_score
#                     else:
#                         recommendations_count[product.strip()] = sim_score

#         # Calcular la probabilidad relativa de cada producto recomendado
#         recommendations_with_prob = []
#         if total_similarity > 0:  # Verificar que total_similarity no sea cero
#             recommendations_with_prob = [(product, score / total_similarity) for product, score in recommendations_count.items()]
#         else:
#             print("No se encontraron similitudes suficientes para calcular probabilidades.")

#         recommendations_with_prob.sort(key=lambda x: x[1], reverse=True)  # Ordenar por puntuación

#         # Crear un nuevo DataFrame para almacenar las recomendaciones con descripciones y probabilidades
#         recommendations_df = pd.DataFrame(columns=['ARTICULO', 'DESCRIPCION', 'PROBABILIDAD'])

#         # Agregar las recomendaciones al DataFrame usando pd.concat
#         for product, prob in recommendations_with_prob:
#             # Buscar la descripción en el DataFrame de productos
#             description = productos.loc[productos['ARTICULO'] == product, 'DESCRIPCION']
#             if not description.empty:
#                 # Crear un nuevo DataFrame temporal para la recomendación
#                 temp_df = pd.DataFrame({
#                     'ARTICULO': [product],
#                     'DESCRIPCION': [description.values[0]],  # Obtener el primer valor encontrado
#                     'PROBABILIDAD': [prob]
#                 })
#                 # Concatenar el DataFrame temporal al DataFrame de recomendaciones
#                 recommendations_df = pd.concat([recommendations_df, temp_df], ignore_index=True)

#         return recommendations_df

#     # Comprobar si el cliente está en el CSV de fieles
#     is_fiel = customer_code in fieles_df['Cliente'].astype(str).values

#     if customer_code:
#         if is_fiel:
#             st.write(f"### Customer {customer_code} is a loyal customer.")
#             option = st.selectbox("Select Recommendation Type", ["Select an option", "By Purchase History", "By Current Basket"])

#             if option == "By Purchase History":
#                 st.warning("Option not available... aún")
#             elif option == "By Current Basket":
#                 st.write("Select the items and assign quantities for the basket:")

#                 # Mostrar lista de artículos disponibles
#                 available_articles = productos['ARTICULO'].unique()
#                 selected_articles = st.multiselect("Select Articles", available_articles)

#                 # Crear inputs para ingresar las cantidades de cada artículo seleccionado
#                 quantities = {}
#                 for article in selected_articles:
#                     quantities[article] = st.number_input(f"Quantity for {article}", min_value=0, step=1)

#                 if st.button("Calcular"):  # Añadimos el botón "Calcular"
#                     # Crear una lista de artículos basada en la selección
#                     new_basket = [f"{article} x{quantities[article]}" for article in selected_articles if quantities[article] > 0]

#                     if new_basket:
#                         # Procesar la lista para recomendar
#                         recommendations_df = recomienda(new_basket)

#                         if not recommendations_df.empty:
#                             st.write("### Recommendations based on the current basket:")
#                             st.dataframe(recommendations_df)
#                         else:
#                             st.warning("No recommendations found for the provided basket.")
#                     else:
#                         st.warning("Please select at least one article and set its quantity.")
#         else:
#             st.write(f"### Customer {customer_code} is not a loyal customer.")
#             st.write("Select items and assign quantities for the basket:")

#             # Mostrar lista de artículos disponibles
#             available_articles = productos['ARTICULO'].unique()
#             selected_articles = st.multiselect("Select Articles", available_articles)

#             # Crear inputs para ingresar las cantidades de cada artículo seleccionado
#             quantities = {}
#             for article in selected_articles:
#                 quantities[article] = st.number_input(f"Quantity for {article}", min_value=0, step=1)

#             if st.button("Calcular"):  # Añadimos el botón "Calcular"
#                 # Crear una lista de artículos basada en la selección
#                 new_basket = [f"{article} x{quantities[article]}" for article in selected_articles if quantities[article] > 0]

#                 if new_basket:
#                     # Procesar la lista para recomendar
#                     recommendations_df = recomienda(new_basket)

#                     if not recommendations_df.empty:
#                         st.write("### Recommendations based on the current basket:")
#                         st.dataframe(recommendations_df)
#                     else:
#                         st.warning("No recommendations found for the provided basket.")
#                 else:
#                     st.warning("Please select at least one article and set its quantity.")