Telecomm Churn Analysis

Project Overview

This project analyzes the different reasons for which customers churn in the telecomm sector. Key themes or factors examined are:

• gender, age
• location : states, cities
• tenure, salary, number of dependents
• data usage

Workflow in this project:

• Step 1 : Exploratory Data Analysis(EDA) using MySQL
• Step 2 : Descriptive and Prescriptive Analysis using Excel
• Step 3 : Visualisations using Power BI

Data Source

• Kaggle Dataset Link: kaggle dataset link
• Actual Dataset copy in inside this repo: source dataset
• The dataset has around 14 columns and 2,43,554 rows

How to Use this Repo

• For obtaining the dataset -> source dataset

• For viewing Exploratory Data Analysis via MySQL : Anurag's exploratory sql analysis

• If unable to import csv file into MySQL Workbench, you can load first 20 rows into a table -> table script with first 20 rows of source data

• To view Excel Analysis (have to download): Anurag Excel Analysis

• To view Power BI reports / dashboards (have to download): Anurag Power BI dashboard

Problem Statement

Univariate Analysis

• Age Groups Demographics : Overview of how our customers are distributed across age groups
• Total Customer Count : The number of customers in the dataset
• Churn Rate : The ratio of customers who churned divided by total customers
• Gender Demographics : Understanding how many of the customers are male and female
• Total Customer Count per Telecomm : Understanding how customers are split between the different telecomms

Bivariate Analysis

• Relationship of Churn with Gender : Impact of Gender on Churning
• Relationship of Churn with Age Groups : Churning across different age groups
• Relationship of Churn with Salary : Analysis of churning across 3 groups - high salary, low salary and normal
• Relationship of Churn with Tenure : Analysing if older customers churn more or new customers churn more
• Relationship of Churn with Number of Dependents : Trend between churning and increase of dependents on the customer
• Relationship of Churn with Number of Data Usage : Examining if providing more data prevents churn
• Churning share across different states : Examining how churning is divided between the states
• Trend of Churn across Cities : Examining which cities have higher churn compared to their counterparts

Immediate Findings

Univariate Analysis Findings

• Age Groups Demographics :

• Total Customer Count & Churn Rate with Telecomms :

• Gender Demographics :

Bivariate Analysis Findings

• Relationship of Churn with Gender :

• Relationship of Churn with Age Groups :

• Relationship of Churn with Salary :

• Relationship of Churn with Tenure :

• Relationship of Churn with Number of Dependents :

• Relationship of Churn with Number of Data Usage :

• Churning share of different states :

• Trend of Churn across Cities :

Key Insights

Inquiry	Actionable Insight
Effect of gender on churning	Among churning customers - males:60% and females:40%
Effect of age groups on Churning	Except the Age Group of 68-77, all age groups have even churning trend
Effect of customer's salary on Churning	Both high and low salary people churn less
Effect of tenure on churning	Depends on the telecomm under study
Effect of number of dependents on churning	Churning is min. if 3, max. if 0
Effect of individual data usage on churning	Depends on the telecomm under study
Effect of states on churning	Evened out share, but maximum churn in Jharkhand
Effect of cities on churning	Hyderabad has max. churn

Visualization