Telecommunications Customer Churn Drivers Analysis

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns


df = pd.read_excel('Telco-Customer-Churn.xlsx')
df.head()

df.info()

## Replacing blanks with 0 as tenure is 0 and no total charges are recorded
df['TotalCharges'] = df['TotalCharges'].replace(" ","0")
df['TotalCharges'] = df['TotalCharges'].astype("float")

df.info()

df.isnull().sum()

df.describe()

df['customerID'].duplicated().sum()

## Converted 0 and 1 values of senior citizens to yes/no to make it easier to understand
def conv(value):
    if value == 1:
        return "yes"
    else:
        return "no"
    

df['SeniorCitizen'] = df['SeniorCitizen'].apply(conv)

df.head(10)

ax = sns.countplot(x='Churn',data=df )
ax.bar_label(ax.containers[0])
plt.title("Count of customers by Churn")
plt.show()

gb = df.groupby("Churn").agg({'Churn':'count'})
plt.title("Percentage of Churned Customers")
plt.pie(gb['Churn'],labels=gb.index,autopct="%1.2f%%")
plt.show()

plt.figure(figsize=(4,4))
ax = sns.countplot(x='gender', data=df, hue="Churn")
plt.title("Churn By Gender")

# Label all bars in all containers
for container in ax.containers:
    ax.bar_label(container)

plt.show()

# Create a crosstab to get counts
ct = pd.crosstab(df['SeniorCitizen'], df['Churn'])

# Calculate percentages
ct_pct = ct.div(ct.sum(axis=1), axis=0) * 100

# Create stacked bar chart
plt.figure(figsize=(4, 4))
ax = ct_pct.plot(kind='bar', stacked=True, color=['#1f77b4', '#ff7f0e'])
plt.title("Churn By Senior Citizen")
plt.xlabel("SeniorCitizen")
plt.ylabel("Percentage (%)")
plt.xticks(rotation=0)
plt.legend(title='Churn')

# Add percentage labels on bars
for container in ax.containers:
    labels = [f'{v:.1f}%' if v > 0 else '' for v in container.datavalues]
    ax.bar_label(container, labels=labels, label_type='center')

plt.tight_layout()
plt.show()

plt.figure(figsize=(6, 4))
ax = sns.countplot(x='SeniorCitizen', data=df, hue='Churn')
plt.title("Churn Count by Senior Citizen Status")
plt.xlabel("Senior Citizen (0=No, 1=Yes)")
plt.ylabel("Count")

# Add count labels on bars
for container in ax.containers:
    ax.bar_label(container)

plt.legend(title='Churn', labels=['No', 'Yes'])
plt.tight_layout()
plt.show()

import matplotlib.pyplot as plt
import pandas as pd

# Normalize labels (supports 0/1 or yes/no)
df['SeniorCitizen'] = df['SeniorCitizen'].replace({0: 'no', 1: 'yes', '0': 'no', '1': 'yes'})
df['Churn'] = df['Churn'].astype(str).str.capitalize()   # "yes"→"Yes"

# Create figure
fig, axes = plt.subplots(2, 2, figsize=(12, 10))
fig.suptitle('Churn Analysis by Senior Citizen Status', fontsize=16, fontweight='bold')

# 1. Grouped Bar Chart (Counts)
ax1 = axes[0, 0]
ct = pd.crosstab(df['SeniorCitizen'], df['Churn'])

ct.plot(kind="bar", ax=ax1, color=['#1f77b4', '#ff7f0e'])
ax1.set_title("Count of Churn by Senior Citizen")
ax1.set_xlabel("Senior Citizen (no/yes)")
ax1.set_ylabel("Count")
ax1.set_xticklabels(ax1.get_xticklabels(), rotation=0)

# Add bar labels
for container in ax1.containers:
    ax1.bar_label(container)

# 2. Stacked Percentage Chart
ax2 = axes[0, 1]
ct_pct = ct.div(ct.sum(axis=1), axis=0) * 100

ct_pct.plot(kind='bar', stacked=True, ax=ax2, color=['#1f77b4', '#ff7f0e'])
ax2.set_title("Churn Rate (%) by Senior Citizen")
ax2.set_xlabel("Senior Citizen (no/yes)")
ax2.set_ylabel("Percentage (%)")
ax2.set_xticklabels(ax2.get_xticklabels(), rotation=0)

# Add percentage labels
for container in ax2.containers:
    labels = [f"{v:.1f}%" if v > 3 else "" for v in container.datavalues]
    ax2.bar_label(container, labels=labels, label_type='center', fontsize=9)


# 3. REMOVE THE EMPTY PLOT (axes[1,0])
axes[1, 0].remove()

# 4. Summary Table
ax4 = axes[1, 1]
ax4.axis('off')

senior_values = df['SeniorCitizen'].unique()
churn_values = df['Churn'].unique()

summary_data = []
for s in senior_values:
    total = len(df[df['SeniorCitizen'] == s])
    for c in churn_values:
        count = len(df[(df['SeniorCitizen'] == s) & (df['Churn'] == c)])
        pct = (count / total * 100) if total else 0
        summary_data.append([f"Senior={s}", c, count, f"{pct:.1f}%"])

table = ax4.table(
    cellText=summary_data,
    colLabels=['Senior Citizen', 'Churn', 'Count', 'Percentage'],
    loc='center',
    cellLoc='center',
    colWidths=[0.25, 0.2, 0.2, 0.25]
)

table.auto_set_font_size(False)
table.set_fontsize(10)
table.scale(1, 2)

# Style header row
for col in range(4):
    table[(0, col)].set_facecolor("#4CAF50")
    table[(0, col)].set_text_props(color="white", weight="bold")

ax4.set_title("Summary Statistics", pad=20, fontweight="bold")

plt.tight_layout()
plt.show()

plt.figure(figsize=(9,5))
sns.histplot(x='tenure',data=df,bins=50, hue='Churn')
plt.show()

plt.figure(figsize=(6, 4))
ax = sns.countplot(x='Contract', data=df, hue='Churn')
plt.title("Churn Count by Contract")

# Add count labels on bars
for container in ax.containers:
    ax.bar_label(container)

plt.legend(title='Churn', labels=['No', 'Yes'])
plt.tight_layout()
plt.show()

df.columns.values

import seaborn as sns
import matplotlib.pyplot as plt
import math


# 1. Identify categorical columns
cat_cols = df.select_dtypes(include=['object']).columns.tolist()

# Remove customerID (unique values, useless for countplot)
if 'customerID' in cat_cols:
    cat_cols.remove('customerID')

# Add SeniorCitizen (0/1)
if 'SeniorCitizen' in df.columns:
    cat_cols.append('SeniorCitizen')

# Remove duplicates if any
cat_cols = list(dict.fromkeys(cat_cols))

# 2. Create subplot grid
n_cols = 3  # number of plots per row
n_rows = math.ceil(len(cat_cols) / n_cols)

plt.figure(figsize=(n_cols * 6, n_rows * 5))  # large grid

# 3. Loop through columns and draw subplots
for i, col in enumerate(cat_cols, 1):
    ax = plt.subplot(n_rows, n_cols, i)
    
    sns.countplot(x=col, data=df, hue='Churn', ax=ax)
    ax.set_title(f"{col} by Churn")
    ax.tick_params(axis='x', rotation=45)

    # Add bar labels
    for container in ax.containers:
        ax.bar_label(container, fontsize=8)

plt.tight_layout()
plt.show()