Exploratory Data
Analysis - Telecom
Churn Analysis
Technical documentation
Mohammed Akifuddin Kashif
[email protected] Introduction
This data set comprises information regarding the churning of a
customer for a Telecom Company. It includes information such as number of
minutes spent on a call, amount charged, total number of calls, whether or
not it was an international call at different times of a day, namely day,
evening(eve), or night. It shows for how long an individual has been loyal to
the company in the form of account length,customer choice on voice plan
and number of customer service calls received.It also includes data
pertaining to customers location i.e, State and Area code.
Problem Statement
We are tasked with performing exploratory data analysis on the given
dataset to get relevant insights from the data and understanding the key
factors responsible for customer churn given particular parameters.
Overview of the data
1. State: Name of the State
2. Account length: Length of the account
3. Area Code:Area where customer resides
4. International Plan: If customer opted for International Plan
5. Voicemail plan: If customer opted for Voicemail Plan
6. Number vmail messages: Number of voicemails received
7. Total day minutes: minutes spent on call during day time
8. Total day calls : Number of calls made during day time
9. Total day charge : Amount charged for calls during day time
10. Total eve minutes: minutes spent on call during evening time
11. Total eve calls :Number of calls made during evening time
12. Total eve charge : Amount charged for calls during evening time
13. Total night minutes : minutes spent on call during night time
14. Total night calls : Number of calls made during night time
15. Total night charge : Amount charged for calls during night time
16. Total night charge : Amount charged for calls during night time
17. Total intl minutes : minutes spent on international calls
18. Total intl calls : Number of calls made internationally
19. Total intl charge : Amount charged for international calls
20. Customer service calls : Number of customer service calls received
21. Churn : If customer churned
Performing EDA (Performing Exploratory Analysis)
Exploratory Data Analysis refers to the critical process of performing initial
investigations on data so as to discover patterns, to spot anomalies, to test
hypotheses and to see assumptions with the assistance of summary
statistics and graphical representations.
1. Extracting head and tail of the dateset
2. Extracting info of the dateset which tells the type of
data present in different columns
3. Producing description of data
4. Checking the unique count of column values.
5. Checked for null values and there weren't any.
6. Creating dummies for data which aren't numeric.
7. Plotting relevant graphs to extract information from
them.
8. Detecting outliers in columns using IQR method and
visualizing them using BoxPlots.
9. Removing them by defining upper and lower limits
using the IQR method.
10. Plotting correlation plots for numeric features to get
additional information
Plots used
Plots are a way to describe the data given to us in a visual manner which is
more understandable and convenient to draw insights from , compared to
raw data. In This project the plots we used are:
1. Countplot - provides the count of required instances from the data.
2. Boxplot - provides information on the outliers present in the data.
3. Histplot - also provides count of instances (seaborn based).
4. Pieplot - provides info in the form of a pie chart.
5. Barplot - BarPlot uses bar to represent the numeric values on the X
and Y axis.
Conclusions from data
● Most of the customers don’t prefer international plans and voicemail
plans.
● Most of the customers spend more time on calls during day time.
● Business is booming in Area Code 415 specifically in the East Coast
Region of the US.
● Customers on average are using the services for a period of around 8
years.
● Most of the customers are located in West Virginia.
● The factors that are affecting the most for customer churn are
International Plan, Total day charge, Customer service calls.
To retain customers :
● The company should give incentives who spend more time on calls
during day time
● The company should concentrate more on prices than features.
● The company should lower prices for international plans and its hourly
call rates.
● The company should decrease the frequency of customer service
calls received by customers.
Challenges Faced
1. Pre-processing the data was one of the challenges we faced, such as detecting
outliers and removing them.
2. Manipulating the data was difficult as it could affect and alter important
information provided by the dataset.
3. As the dataset has more columns it was quite difficult to extract relevant
information from all the columns.
Final Conclusion
● We are finally at the conclusion of our project!
● We performed exploratory data analysis from the given dataset and
were able to draw important and relevant information which explains
why in a particular instance there are more customers who were loyal
or have churned compared to other instances.
● We also used different types of plots to better visualize and present
the data in a more appealing manner.
