
import pandas as pd
import numpy as np
from datetime import datetime
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.preprocessing import LabelEncoder, StandardScaler
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from yellowbrick.cluster import KElbowVisualizer
from sklearn.metrics import silhouette_score, calinski_harabasz_score, davies_bouldin_score
from math import pi
import matplotlib.patches as mpatches


# Load the dataset
file_path = '../data/marketing_campaign.csv'
df = pd.read_csv(file_path, sep='\t')
pd.set_option('display.max_columns', None)

# Rename purchase amount columns for clarity
df.rename(columns=lambda x: x.replace('Mnt', 'Amount') if x.startswith('Mnt') else x, inplace=True)

# Display the first few rows of the dataset
df.head()


# Checking for missing values
missing_values = df.isnull().sum()

# Checking data types and unique values for categorical features
data_types = df.dtypes
unique_values = {col: df[col].unique() for col in df.select_dtypes(include=['object', 'category']).columns}

missing_values, data_types, unique_values


# Dropping rows with 'Absurd', 'Alone', and 'YOLO' in Marital Status before encoding
df = df[~df['Marital_Status'].isin(['Absurd', 'Alone', 'YOLO'])]

# Handling missing values in 'Income'
df['Income'].fillna(df['Income'].median(), inplace=True)

# Converting 'Dt_Customer' to datetime format
df['Dt_Customer'] = pd.to_datetime(df['Dt_Customer'], format='%d-%m-%Y')

# Ensuring that categorical variables are encoded into separate columns (One-hot encoding)
df = pd.get_dummies(df, columns=['Education', 'Marital_Status'], prefix=['education', 'marital_status'], drop_first=True)

# Removing rows with negative income values and capping income at the upper bound
df = df[df['Income'] >= 0]
df = df.drop(df[df['ID'] == 9432].index) # Drop a record with an invalid income

# Calculate Age
df['Age'] = 2024 - df['Year_Birth']
df['Age'] = df['Age'].clip(upper=86)

# Displaying the cleaned dataset's first few rows to confirm the preprocessing steps
df.head()


# Setting the seaborn style to 'darkgrid'
sns.set_style("darkgrid")


# Plotting Age Distribution
plt.figure(figsize=(8, 6))
sns.histplot(df['Age'], bins=20, kde=False, color='skyblue', edgecolor='black')
plt.title('Distribution of Age')
plt.xlabel('Age')
plt.ylabel('Number of Customers')
plt.show()


# Plotting Income Distribution
plt.figure(figsize=(8, 6))
sns.histplot(df['Income'], bins=20, kde=False, color='lightgreen', edgecolor='black')
plt.title('Distribution of Income')
plt.xlabel('Income')
plt.ylabel('Number of Customers')
plt.show()


# Distribution of Education Levels
plt.figure(figsize=(8, 6))
education_columns = [col for col in df.columns if 'education_' in col]
education_counts = df[education_columns].sum()
education_counts.plot(kind='bar', color='lightcoral', edgecolor='black')
plt.title('Distribution of Education Levels')
plt.xlabel('Education Level')
plt.ylabel('Number of Customers')
plt.xticks(ticks=range(len(education_columns)), labels=[col.split('_')[1] for col in education_columns], rotation=45)
plt.show()


# Distribution of Marital Status
plt.figure(figsize=(8, 6))
marital_status_columns = [col for col in df.columns if 'marital_status_' in col]
marital_status_counts = df[marital_status_columns].sum()
marital_status_counts.plot(kind='bar', color='lightblue', edgecolor='black')
plt.title('Distribution of Marital Status')
plt.xlabel('Marital Status')
plt.ylabel('Number of Customers')
plt.xticks(ticks=range(len(marital_status_columns)), labels=[col.split('_')[2] for col in marital_status_columns], rotation=45)
plt.show()


# List of product categories
product_categories = ['AmountWines', 'AmountFruits', 'AmountMeatProducts', 'AmountFishProducts', 'AmountSweetProducts', 'AmountGoldProds']

# Plotting spending behavior on various product categories on the same figure
plt.figure(figsize=(14, 10))

for i, category in enumerate(product_categories, 1):
    plt.subplot(3, 2, i)
    sns.histplot(df[category], bins=20, kde=False, color='lightcoral', edgecolor='black')
    plt.title(f'Spending on {category}')
    plt.xlabel('Amount Spent')
    plt.ylabel('Number of Customers')

plt.tight_layout()
plt.show()


# Analyzing purchase channels: Web, Catalog, and Store purchases

# List of purchase channel columns
purchase_channels = ['NumWebPurchases', 'NumCatalogPurchases', 'NumStorePurchases']

# Plotting distribution of purchases across different channels
plt.figure(figsize=(14, 6))

for i, channel in enumerate(purchase_channels, 1):
    plt.subplot(1, 3, i)
    sns.histplot(df[channel], bins=20, kde=False, color='lightblue', edgecolor='black')
    plt.title(f'Distribution of {channel}')
    plt.xlabel('Number of Purchases')
    plt.ylabel('Number of Customers')

plt.tight_layout()
plt.show()


# List of campaign response columns
campaign_columns = ['AcceptedCmp1', 'AcceptedCmp2', 'AcceptedCmp3', 'AcceptedCmp4', 'AcceptedCmp5', 'Response']
campaign_labels = ['Campaign 1', 'Campaign 2', 'Campaign 3', 'Campaign 4', 'Campaign 5', 'Last Campaign']

# Calculating the acceptance rates for each campaign
campaign_data = df[campaign_columns].sum()

# Plotting the acceptance rates for each campaign
plt.figure(figsize=(10, 6))
sns.barplot(x=campaign_labels, y=campaign_data.values, palette="viridis")
plt.title('Acceptance Rates for Different Campaigns')
plt.xlabel('Campaigns')
plt.ylabel('Number of Acceptances')
plt.show()


# Selecting relevant features for segmentation
features = ['Age', 'Income', 'AmountWines', 'AmountFruits', 'AmountMeatProducts',
            'AmountFishProducts', 'AmountSweetProducts', 'AmountGoldProds'] + \
           [col for col in df.columns if 'education_' in col] + \
           [col for col in df.columns if 'marital_status_' in col]

# Extracting the features data
X = df[features]

# Normalizing the data
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# Determining the optimal number of clusters using the elbow method
inertia = []
for n_clusters in range(1, 11):
    kmeans = KMeans(n_clusters=n_clusters, random_state=42)
    kmeans.fit(X_scaled)
    inertia.append(kmeans.inertia_)

# Plotting the elbow plot
plt.figure(figsize=(8, 6))
plt.plot(range(1, 11), inertia, marker='o')
plt.title('Elbow Plot for Determining Optimal Number of Clusters')
plt.xlabel('Number of Clusters')
plt.ylabel('Inertia')
plt.grid(True)
plt.show()

c:\ProgramData\Anaconda3\lib\site-packages\sklearn\cluster\_kmeans.py:1036: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=9.
  warnings.warn(


# Performing K-means clustering with 5 clusters
kmeans = KMeans(n_clusters=2, random_state=42)
df['Cluster'] = kmeans.fit_predict(X_scaled)

# Analyzing the characteristics of each cluster
cluster_centers = scaler.inverse_transform(kmeans.cluster_centers_)
cluster_centers_df = pd.DataFrame(cluster_centers, columns=features)
cluster_centers_df['Cluster'] = range(1, 3)

# Displaying the cluster centers to understand the characteristics of each cluster
cluster_centers_df


# Creating a scatter plot to visualize clusters based on Income and Total Spent
# Assuming 'TotalSpent' is a calculated column for the total spending
df['TotalSpent'] = df['AmountWines'] + df['AmountFruits'] + df['AmountMeatProducts'] + df['AmountFishProducts'] + df['AmountSweetProducts'] + df['AmountGoldProds']

# Creating the scatter plot
plt.figure(figsize=(10, 6))
pl = sns.scatterplot(data=df, x='TotalSpent', y='Income', hue='Cluster', palette='viridis', alpha=0.6)
pl.set_title("Cluster's Profile Based On Income And Spending")
plt.xlabel('Total Spent')
plt.ylabel('Income')
plt.legend(title='Cluster')
plt.grid(True)
plt.show()


# Ensure 'df' has only numeric columns and excludes non-numeric columns like timestamps
numeric_df = df.select_dtypes(include=['number'])

# Extract features and labels
X = numeric_df.drop(columns=['Cluster'])  # Exclude 'Cluster' column if present in the features
y = df['Cluster']  # Cluster labels

# Calculate the silhouette score
silhouette_avg = silhouette_score(X, y)
print(f'Silhouette Score: {silhouette_avg}')

# Calculate the Calinski-Harabasz Index
calinski_harabasz = calinski_harabasz_score(X, y)
print(f'Calinski-Harabasz Index: {calinski_harabasz}')

# Calculate the Davies-Bouldin Index
davies_bouldin = davies_bouldin_score(X, y)
print(f'Davies-Bouldin Index: {davies_bouldin}')

Silhouette Score: 0.4303628272179567
Calinski-Harabasz Index: 2297.901822209848
Davies-Bouldin Index: 0.718969750881457


# Assuming 'Cluster' is the column with the main cluster labels and 'df' is your DataFrame
cluster_0_data = df[df['Cluster'] == 0]
cluster_1_data = df[df['Cluster'] == 1]

# Convert datetime columns if necessary, or drop them if they are not useful for clustering
# For example, converting 'Dt_Customer' to the number of days since a specific date
df['Days_Since_Enrollment'] = (pd.Timestamp('now') - pd.to_datetime(df['Dt_Customer'])).dt.days

# Drop non-numeric columns and the original datetime column
df_numeric = df.drop(columns=['Cluster', 'Dt_Customer'])  # Add other non-numeric columns if necessary

# Filter only numeric data
numeric_columns = df_numeric.select_dtypes(include=['number']).columns
X_cluster_0 = df[df['Cluster'] == 0][numeric_columns]
X_cluster_1 = df[df['Cluster'] == 1][numeric_columns]

# Apply K-Means to find sub-clusters within cluster 0
kmeans_cluster_0 = KMeans(n_clusters=2, random_state=42).fit(X_cluster_0)
df.loc[df['Cluster'] == 0, 'Subcluster'] = kmeans_cluster_0.labels_

# Apply K-Means to find sub-clusters within cluster 1
kmeans_cluster_1 = KMeans(n_clusters=2, random_state=42).fit(X_cluster_1)
df.loc[df['Cluster'] == 1, 'Subcluster'] = kmeans_cluster_1.labels_


# For sub-clusters within the original cluster 0
silhouette_avg_0 = silhouette_score(X_cluster_0, df[df['Cluster'] == 0]['Subcluster'])
calinski_harabasz_0 = calinski_harabasz_score(X_cluster_0, df[df['Cluster'] == 0]['Subcluster'])
davies_bouldin_0 = davies_bouldin_score(X_cluster_0, df[df['Cluster'] == 0]['Subcluster'])

print(f'Sub-Cluster 0 - Silhouette Score: {silhouette_avg_0}')
print(f'Sub-Cluster 0 - Calinski-Harabasz Index: {calinski_harabasz_0}')
print(f'Sub-Cluster 0 - Davies-Bouldin Index: {davies_bouldin_0}')

# For sub-clusters within the original cluster 1
silhouette_avg_1 = silhouette_score(X_cluster_1, df[df['Cluster'] == 1]['Subcluster'])
calinski_harabasz_1 = calinski_harabasz_score(X_cluster_1, df[df['Cluster'] == 1]['Subcluster'])
davies_bouldin_1 = davies_bouldin_score(X_cluster_1, df[df['Cluster'] == 1]['Subcluster'])

print(f'Sub-Cluster 1 - Silhouette Score: {silhouette_avg_1}')
print(f'Sub-Cluster 1 - Calinski-Harabasz Index: {calinski_harabasz_1}')
print(f'Sub-Cluster 1 - Davies-Bouldin Index: {davies_bouldin_1}')

Sub-Cluster 0 - Silhouette Score: 0.5255779363538116
Sub-Cluster 0 - Calinski-Harabasz Index: 2170.9155462471986
Sub-Cluster 0 - Davies-Bouldin Index: 0.6508987002199438
Sub-Cluster 1 - Silhouette Score: 0.46130052771548885
Sub-Cluster 1 - Calinski-Harabasz Index: 693.6154191301382
Sub-Cluster 1 - Davies-Bouldin Index: 0.760878426497644


# Ensure 'Total_Spent' is calculated
if 'Total_Spent' not in df.columns:
    df['Total_Spent'] = df[['AmountWines', 'AmountFruits', 'AmountMeatProducts', 'AmountFishProducts', 'AmountSweetProducts', 'AmountGoldProds']].sum(axis=1)

# Convert datetime columns if necessary, or drop them if they are not useful for clustering
if 'Dt_Customer' in df.columns:
    df['Days_Since_Enrollment'] = (pd.Timestamp('now') - pd.to_datetime(df['Dt_Customer'])).dt.days

# Drop non-numeric columns and the original datetime column
df_numeric = df.select_dtypes(include=[np.number]).drop(columns=['Cluster', 'Subcluster'], errors='ignore')


# Subclustering Process
for main_cluster in df['Cluster'].unique():
    main_cluster_data = df[df['Cluster'] == main_cluster]
    if main_cluster_data.empty:
        continue
    
    X = main_cluster_data[df_numeric.columns]
    kmeans = KMeans(n_clusters=2, random_state=42)
    subcluster_labels = kmeans.fit_predict(X)
    df.loc[df['Cluster'] == main_cluster, 'Subcluster'] = subcluster_labels + main_cluster * 2

# Verify that Subcluster column has a wider range of labels
print(f"Subclusters: {df['Subcluster'].unique()}")

Subclusters: [2. 1. 0. 3.]


# Mapping subcluster numbers to their descriptive names
df['SubclusterName'] = df['Subcluster'].map(subcluster_names)

# Creating a scatter plot to visualize subclusters based on Income and Total Spent
plt.figure(figsize=(10, 6))
pl = sns.scatterplot(data=df, x='TotalSpent', y='Income', hue='SubclusterName', palette='viridis', alpha=0.6)
pl.set_title("Subcluster's Profile Based On Income And Spending")
plt.xlabel('Total Spent')
plt.ylabel('Income')
plt.legend(title='Subcluster')
plt.grid(True)
plt.show()


# Ensure 'TotalSpent' is calculated if not already done
if 'TotalSpent' not in df.columns:
    df['TotalSpent'] = df[['AmountWines', 'AmountFruits', 'AmountMeatProducts', 'AmountFishProducts', 'AmountSweetProducts', 'AmountGoldProds']].sum(axis=1)

# Mapping subcluster numbers to their descriptive names
df['SubclusterName'] = df['Subcluster'].map(subcluster_names)

# Selecting key features for the pairplot
pairplot_features = ['Age', 'Income', 'TotalSpent', 'NumWebPurchases', 'NumCatalogPurchases', 'NumStorePurchases', 'SubclusterName']

# Creating the pairplot
sns.pairplot(df[pairplot_features], hue='SubclusterName', palette='viridis', 
             plot_kws={'alpha':0.4}, diag_kws={'alpha':0.4})
plt.suptitle('Pairplot of Subclusters by Key Features', y=1.02)
plt.show()


# Adjusted profiling with rounding purchases to two decimal points
subcluster_profiles = []

# Assuming 'Children', 'Complain', and discount-related columns exist, along with marital status and education columns
marital_status_cols = ['marital_status_Married', 'marital_status_Single', 'marital_status_Together', 'marital_status_Widow']
education_cols = ['education_Basic', 'education_Graduation', 'education_Master', 'education_PhD']
discount_cols = ['NumDealsPurchases', 'NumWebPurchases', 'NumCatalogPurchases', 'NumStorePurchases']

# Assigning meaningful names to each subcluster
subcluster_names = {
    0.0: 'Working Class',
    1.0: 'Emerging Professionals',
    2.0: 'Value-Oriented Professionals',
    3.0: 'High-Income Luxury Seekers'
}

for subcluster in df['Subcluster'].unique():
    # Filter data for the current subcluster
    subcluster_data = df[df['Subcluster'] == subcluster]
    
    # Demographic Profiling
    age_summary = subcluster_data['Age'].describe()
    income_summary = subcluster_data['Income'].describe()
    
    # Behavioral Profiling
    total_spent_summary = subcluster_data['Total_Spent'].describe()
    purchasing_patterns = subcluster_data[['AmountWines', 'AmountFruits', 'AmountMeatProducts', 'AmountFishProducts', 'AmountSweetProducts', 'AmountGoldProds']].mean()
    channel_preferences = subcluster_data[['NumWebPurchases', 'NumCatalogPurchases', 'NumStorePurchases']].mean()
    
    # Additional Features
    children_avg = subcluster_data['Kidhome'].mean() + subcluster_data['Teenhome'].mean()
    complaints_percentage = subcluster_data['Complain'].mean() * 100
    marital_status_percentage = (subcluster_data[marital_status_cols].mean() * 100).to_dict()
    education_percentage = (subcluster_data[education_cols].mean() * 100).to_dict()
    
    # Calculate overall discount percentage
    total_discount_purchases = subcluster_data[discount_cols].sum(axis=1).sum()
    total_spent = subcluster_data['Total_Spent'].sum()
    discounts_percentage = (total_discount_purchases / total_spent) * 100 if total_spent != 0 else 0
    
    # Creating a profile dictionary for the current subcluster with formatted values
    profile = {
        'Subcluster': subcluster,
        'Name': subcluster_names.get(subcluster, 'Unknown'),
        'Age_Mean': round(age_summary['mean']),
        'Age_Median': age_summary['50%'],
        'Age_Q1': age_summary['25%'],
        'Age_Q3': age_summary['75%'],
        'Income_Mean': f"${income_summary['mean']:,.0f}",
        'Income_Median': f"${income_summary['50%']:,.0f}",
        'Income_Q1': f"${income_summary['25%']:,.0f}",
        'Income_Q3': f"${income_summary['75%']:,.0f}",
        'Total_Spent_Mean': f"${total_spent_summary['mean']:,.0f}",
        'Total_Spent_Median': f"${total_spent_summary['50%']:,.0f}",
        'Purchases_Wines': round(purchasing_patterns['AmountWines'], 2),
        'Purchases_Fruits': round(purchasing_patterns['AmountFruits'], 2),
        'Purchases_Meat': round(purchasing_patterns['AmountMeatProducts'], 2),
        'Purchases_Fish': round(purchasing_patterns['AmountFishProducts'], 2),
        'Purchases_Sweets': round(purchasing_patterns['AmountSweetProducts'], 2),
        'Purchases_Gold': round(purchasing_patterns['AmountGoldProds'], 2),
        'NumWebPurchases': f"{channel_preferences['NumWebPurchases']:.1f}",
        'NumCatalogPurchases': f"{channel_preferences['NumCatalogPurchases']:.1f}",
        'NumStorePurchases': f"{channel_preferences['NumStorePurchases']:.1f}",
        'Children_Avg': f"{children_avg:.2f}",
        'Complaints_%': f"{complaints_percentage:.2f}%",
        'Discounts_%': f"{discounts_percentage:.2f}%"
    }
    
    # Adding formatted marital status and education percentages
    profile.update({f"Marital_Status_{status}_%": f"{percentage:.2f}%" for status, percentage in marital_status_percentage.items()})
    profile.update({f"Education_{edu_level}_%": f"{percentage:.2f}%" for edu_level, percentage in education_percentage.items()})
    
    # Append the profile data to the list
    subcluster_profiles.append(profile)

# Convert the list of profiles into a DataFrame for better visualization
subcluster_profiles_df = pd.DataFrame(subcluster_profiles)

# Display the profiles in a table format
subcluster_profiles_df


# Preparing the data for plotting
marital_status_df = pd.DataFrame({
    'Subcluster': df['Subcluster'],
    'Married': df['marital_status_Married'],
    'Single': df['marital_status_Single'],
    'Together': df['marital_status_Together'],
    'Widow': df['marital_status_Widow']
})

# Melting the dataframe for seaborn
marital_status_df = marital_status_df.melt(id_vars='Subcluster', var_name='Marital_Status', value_name='Proportion')

# Mapping subcluster numbers to their descriptive names
marital_status_df['SubclusterName'] = marital_status_df['Subcluster'].map(subcluster_names)

# Plotting the side-by-side bar chart without error bars
plt.figure(figsize=(14, 8))
sns.barplot(data=marital_status_df, x='SubclusterName', y='Proportion', hue='Marital_Status', palette='viridis', ci=None)
plt.title('Marital Status Distribution Across Subclusters')
plt.xlabel('Subcluster')
plt.ylabel('Proportion')
plt.xticks(fontsize=12)
plt.legend(title='Marital Status')
plt.grid(True)
plt.show()


# Aggregating the data to get average spending in each category per subcluster
average_spending = df.groupby('Subcluster').agg({
    'AmountWines': 'mean',
    'AmountFruits': 'mean',
    'AmountMeatProducts': 'mean',
    'AmountFishProducts': 'mean',
    'AmountSweetProducts': 'mean',
    'AmountGoldProds': 'mean'
}).reset_index()

# Melting the DataFrame for seaborn
average_spending_melted = average_spending.melt(id_vars='Subcluster', var_name='Product_Category', value_name='Average_Spent')

# Removing the 'Amount' prefix from Product_Category
average_spending_melted['Product_Category'] = average_spending_melted['Product_Category'].str.replace('Amount', '')

# Calculate the proportion of spending for each product category within each subcluster
average_spending_melted['Proportion_Spent'] = average_spending_melted.groupby('Subcluster')['Average_Spent'].transform(lambda x: x / x.sum())

# Create a distribution DataFrame using crosstab
distribution = pd.crosstab(average_spending_melted.Subcluster, average_spending_melted.Product_Category, values=average_spending_melted.Proportion_Spent, aggfunc='mean', normalize='index').fillna(0)

# Define the viridis color palette
viridis_palette = sns.color_palette("viridis", len(distribution.columns))

# Map subcluster numbers to their descriptive names
distribution.index = distribution.index.map(subcluster_names)

# Plotting the cumsum with reverse hue order and viridis color palette
plt.figure(figsize=(11, 10))  # Increased figure size
sns.barplot(data=distribution.cumsum(axis=1).stack().reset_index(name='Dist'),
            x='Subcluster', y='Dist', hue='Product_Category',
            hue_order=distribution.columns[::-1], dodge=False, palette=viridis_palette)
plt.title('Proportion of Purchases Across Product Categories by Subcluster')
plt.xlabel('Subcluster')
plt.ylabel('Proportion Spent')
plt.xticks(ticks=range(len(distribution)), labels=distribution.index)
plt.legend(title='Product Category', bbox_to_anchor=(1.05, 1), loc='upper left')
plt.grid(True)
plt.show()


# Categories for the radar chart without 'Amount' prefix
categories = ['Wines', 'Fruits', 'Meat Products', 'Fish Products', 'Sweet Products', 'Gold Products']
N = len(categories)

# Calculating angles for each category axis
angles = [n / float(N) * 2 * pi for n in range(N)]
angles += angles[:1]  # Complete the loop

# Define the appropriate scaling for the data
max_value = average_spending[['AmountWines', 'AmountFruits', 'AmountMeatProducts', 'AmountFishProducts', 'AmountSweetProducts', 'AmountGoldProds']].max().max()
max_range = int(max_value // 100 + 1) * 100  # Round up to the nearest hundred for a nice axis range

# Initialize the radar chart with a zoomed-in scale
plt.figure(figsize=(10, 8))
ax = plt.subplot(111, polar=True)

# Define a color palette
colors = ['b', 'g', 'r', 'm']

# Plotting data for each subcluster
for i in range(len(average_spending)):
    values = average_spending.loc[i, ['AmountWines', 'AmountFruits', 'AmountMeatProducts', 'AmountFishProducts', 'AmountSweetProducts', 'AmountGoldProds']].tolist()
    values += values[:1]  # Repeat the first value to close the circle
    subcluster_label = subcluster_names.get(average_spending["Subcluster"][i], f'Subcluster {int(average_spending["Subcluster"][i])}')
    ax.plot(angles, values, linewidth=2, linestyle='solid', label=subcluster_label, color=colors[i])
    ax.fill(angles, values, color=colors[i], alpha=0.25)

# Add labels for each category without the 'Amount' prefix
plt.xticks(angles[:-1], categories, size=12)
plt.yticks(np.linspace(0, max_range, num=5), [f"{int(x)}" for x in np.linspace(0, max_range, num=5)], size=10)

# Title and legend
plt.title('Average Purchases Across Product Categories by Subcluster')
plt.legend(loc='upper right', bbox_to_anchor=(1.3, 1.1))

# Show the plot
plt.show()


# Define the adjusted names for the subclusters
subcluster_names = {
    0: 'Working Class',
    1: 'Emerging Professionals',
    2: 'Value-Oriented Professionals',
    3: 'High-Income Luxury Seekers'
}

# Function to plot KDE plots for Age for all subclusters with custom names and colors
def plot_age_kde(df, subcluster_names):
    plt.figure(figsize=(12, 8))
    colors = sns.color_palette("hsv", len(subcluster_names))
    for subcluster, color in zip(df['Subcluster'].unique(), colors):
        subcluster_data = df[df['Subcluster'] == subcluster]
        sns.kdeplot(subcluster_data['Age'], label=subcluster_names[subcluster], color=color, shade=True, linewidth=2)
    plt.title('Age Distributions Across Subclusters')
    plt.xlabel('Age')
    plt.ylabel('Density')
    plt.legend(title='Subcluster')
    plt.grid(True)
    plt.show()

# Function to plot KDE plots for Income for all subclusters with custom names and colors
def plot_income_kde(df, subcluster_names):
    plt.figure(figsize=(12, 8))
    colors = sns.color_palette("hsv", len(subcluster_names))
    for subcluster, color in zip(df['Subcluster'].unique(), colors):
        subcluster_data = df[df['Subcluster'] == subcluster]
        sns.kdeplot(subcluster_data['Income'], label=subcluster_names[subcluster], color=color, shade=True, linewidth=2)
    plt.title('Income Distributions Across Subclusters')
    plt.xlabel('Income')
    plt.ylabel('Density')
    plt.legend(title='Subcluster')
    plt.grid(True)
    plt.show()

    # Function to plot KDE plots for Total Spending for all subclusters with custom names and colors
def plot_total_spent_kde(df, subcluster_names):
    plt.figure(figsize=(12, 8))
    colors = sns.color_palette("hsv", len(subcluster_names))
    for subcluster, color in zip(df['Subcluster'].unique(), colors):
        subcluster_data = df[df['Subcluster'] == subcluster]
        sns.kdeplot(subcluster_data['Total_Spent'], label=subcluster_names[subcluster], color=color, shade=True, linewidth=2)
    plt.title('Total Spending Distributions Across Subclusters')
    plt.xlabel('Total Spent')
    plt.ylabel('Density')
    plt.legend(title='Subcluster')
    plt.grid(True)
    plt.show()

# Plotting KDE plots for Age, Income, and Total Spending
plot_age_kde(df, subcluster_names)
plot_income_kde(df, subcluster_names)
plot_total_spent_kde(df, subcluster_names)


# Define the adjusted names for the subclusters
subcluster_names = {
    0: 'Working Class',
    1: 'Emerging Professionals',
    2: 'Value-Oriented Professionals',
    3: 'High-Income Luxury Seekers'
}

# Function to plot box plots for Age for all subclusters
def plot_age_boxplot(df, subcluster_names):
    plt.figure(figsize=(12, 8))
    sns.boxplot(x='Subcluster', y='Age', data=df, palette='viridis')
    plt.title('Age Distribution Across Subclusters')
    plt.xlabel('Subcluster')
    plt.ylabel('Age')
    plt.xticks(ticks=[0, 1, 2, 3], labels=[subcluster_names[0], subcluster_names[1], subcluster_names[2], subcluster_names[3]])
    plt.grid(True)
    plt.show()

# Function to plot box plots for Income for all subclusters
def plot_income_boxplot(df, subcluster_names):
    plt.figure(figsize=(12, 8))
    sns.boxplot(x='Subcluster', y='Income', data=df, palette='viridis')
    plt.title('Income Distribution Across Subclusters')
    plt.xlabel('Subcluster')
    plt.ylabel('Income')
    plt.xticks(ticks=[0, 1, 2, 3], labels=[subcluster_names[0], subcluster_names[1], subcluster_names[2], subcluster_names[3]])
    plt.grid(True)
    plt.show()

# Plotting box plots for Age and Income
plot_age_boxplot(df, subcluster_names)
plot_income_boxplot(df, subcluster_names)


# Preparing the data for plotting
campaign_columns = ['AcceptedCmp1', 'AcceptedCmp2', 'AcceptedCmp3', 'AcceptedCmp4', 'AcceptedCmp5']
df_campaign = df.groupby('Subcluster')[campaign_columns].mean().reset_index()

# Melting the dataframe for seaborn
df_campaign_melted = df_campaign.melt(id_vars='Subcluster', var_name='Campaign', value_name='Acceptance_Rate')

# Renaming campaigns to show just the number
df_campaign_melted['Campaign'] = df_campaign_melted['Campaign'].apply(lambda x: x[-1])

# Map subcluster numbers to their descriptive names
df_campaign_melted['Subcluster'] = df_campaign_melted['Subcluster'].map(subcluster_names)

# Plotting the stacked bar chart
plt.figure(figsize=(10, 6))
sns.barplot(data=df_campaign_melted, x='Subcluster', y='Acceptance_Rate', hue='Campaign', palette='viridis')
plt.title('Campaign Acceptance Rate by Subcluster')
plt.xlabel('Subcluster')
plt.ylabel('Acceptance Rate')
plt.legend(title='Campaign')
plt.grid(True)
plt.tight_layout()
plt.show()


# Calculating mean acceptance rates for each campaign and cluster
df_campaign_mean = df.groupby('Subcluster')[campaign_columns].mean()

# Mapping subcluster numbers to their descriptive names
df_campaign_mean.index = df_campaign_mean.index.map(subcluster_names)

# Renaming the columns to just show the campaign numbers
df_campaign_mean.columns = [str(i + 1) for i in range(len(df_campaign_mean.columns))]

# Plotting the heatmap with the 'coolwarm' colormap
plt.figure(figsize=(10, 6))
sns.heatmap(df_campaign_mean, annot=True, cmap='plasma', fmt=".2f")
plt.title('Campaign Acceptance Rates by Subcluster')
plt.xlabel('Campaign')
plt.ylabel('Subcluster')
plt.tight_layout()
plt.show()

	Age	Income	AmountWines	AmountFruits	AmountMeatProducts	AmountFishProducts	AmountSweetProducts	AmountGoldProds	education_Basic	education_Graduation	education_Master	education_PhD	marital_status_Married	marital_status_Single	marital_status_Together	marital_status_Widow	Cluster
0	54.323003	40648.263774	140.750000	7.313361	44.668733	10.437328	7.222452	24.655647	0.036501	0.476584	0.179063	0.216942	0.396006	0.215565	0.261019	0.026860	1
1	56.712821	73015.848718	608.370513	61.662821	395.239744	87.758974	64.160256	79.774359	0.001282	0.553846	0.138462	0.215385	0.370513	0.214103	0.256410	0.048718	2

	Subcluster	Name	Age_Mean	Age_Median	Age_Q1	Age_Q3	Income_Mean	Income_Median	Income_Q1	Income_Q3	Total_Spent_Mean	Total_Spent_Median	Purchases_Wines	Purchases_Fruits	Purchases_Meat	Purchases_Fish	Purchases_Sweets	Purchases_Gold	NumWebPurchases	NumCatalogPurchases	NumStorePurchases	Children_Avg	Complaints_%	Discounts_%	Marital_Status_marital_status_Married_%	Marital_Status_marital_status_Single_%	Marital_Status_marital_status_Together_%	Marital_Status_marital_status_Widow_%	Education_education_Basic_%	Education_education_Graduation_%	Education_education_Master_%	Education_education_PhD_%
0	2.0	Value-Oriented Professionals	57	57.0	48.0	67.00	$63,965	$65,832	$60,924	$70,069	$1,143	$1,092	548.27	57.24	310.56	81.28	58.93	86.68	6.0	4.9	8.7	0.60	0.50%	1.94%	36.25%	20.25%	24.75%	6.00%	0.25%	57.75%	11.25%	21.25%
1	1.0	Emerging Professionals	58	57.5	50.0	67.00	$53,705	$51,742	$46,380	$58,654	$393	$338	260.00	9.53	68.91	12.88	9.11	32.51	4.5	1.8	5.7	1.30	0.58%	3.81%	40.00%	19.57%	25.80%	3.62%	0.00%	43.91%	21.59%	27.39%
2	0.0	Working Class	51	50.0	43.0	57.00	$28,825	$30,360	$23,007	$35,889	$92	$57	32.77	5.30	22.72	8.22	5.52	17.54	2.1	0.5	3.1	1.14	1.71%	8.54%	39.24%	23.36%	26.38%	1.84%	6.96%	51.05%	14.57%	16.54%
3	3.0	High-Income Luxury Seekers	56	55.0	46.0	66.25	$82,543	$80,690	$77,042	$84,842	$1,459	$1,454	671.64	66.32	484.38	94.58	69.67	72.51	5.3	6.2	8.3	0.29	0.53%	1.44%	37.89%	22.63%	26.58%	3.68%	0.00%	52.89%	16.58%	21.84%

Customer Segmentation Analysis¶

Loading Data¶

Data Preprocessing¶

Exploratory Data Analysis¶

Demographics¶

Spending¶

Purchase Channels¶

Campaign Response¶

Cluster Analysis¶

Segmentation¶

Cluster Analysis¶

	ID	Year_Birth	Education	Marital_Status	Income	Kidhome	Teenhome	Dt_Customer	Recency	AmountWines	AmountFruits	AmountMeatProducts	AmountFishProducts	AmountSweetProducts	AmountGoldProds	NumDealsPurchases	NumWebPurchases	NumCatalogPurchases	NumStorePurchases	NumWebVisitsMonth	Z_CostContact	Z_Revenue	Response
0	5524	1957	Graduation	Single	58138.0	0	0	04-09-2012	58	635	88	546	172	88	88	3	8	10	4	7	3	11	1
1	2174	1954	Graduation	Single	46344.0	1	1	08-03-2014	38	11	1	6	2	1	6	2	1	1	2	5	3	11	0
2	4141	1965	Graduation	Together	71613.0	0	0	21-08-2013	26	426	49	127	111	21	42	1	8	2	10	4	3	11	0
3	6182	1984	Graduation	Together	26646.0	1	0	10-02-2014	26	11	4	20	10	3	5	2	2	0	4	6	3	11	0
4	5324	1981	PhD	Married	58293.0	1	0	19-01-2014	94	173	43	118	46	27	15	5	5	3	6	5	3	11	0