Scribd
Upload
16 views

1ijecs

The paper analyzes customer satisfaction in the telecommunications sector using big data from social media, specifically tweets about Verizon. It employs a random forest classifier to categorize customer sentiments into positive, negative, and neutral, following extensive data preprocessing and exploratory analysis. The findings suggest that improving service quality and utilizing real-time data can enhance customer satisfaction and reduce churn.

Uploaded by

Olfa Trabelsi
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
16 views

1ijecs

The paper analyzes customer satisfaction in the telecommunications sector using big data from social media, specifically tweets about Verizon. It employs a random forest classifier to categorize customer sentiments into positive, negative, and neutral, following extensive data preprocessing and exploratory analysis. The findings suggest that improving service quality and utilizing real-time data can enhance customer satisfaction and reduce churn.

Uploaded by

Olfa Trabelsi
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/375592850

Analyzing Customer Satisfaction Level on Telecommunication Usage Using Big


Data

Article · September 2023


DOI: 10.18535/ijecs/v12i09.4757

CITATIONS READS
0 159

3 authors, including:

Justice Palimote Chizi Aloy-Okwelle


kenule Beeson Saro-wiwa polytechnic, Bori, Rivers State. Rivers State University
12 PUBLICATIONS 12 CITATIONS 3 PUBLICATIONS 0 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Chizi Aloy-Okwelle on 12 November 2023.

The user has requested enhancement of the downloaded file.


www.ijecs.in
International Journal Of Engineering And Computer Science
Volume 12 Issue 09 September 2023, Page No. 25801-25808
ISSN: 2319-7242 DOI: 10.18535/ijecs/v12i09.4757

Analyzing Customer Satisfaction Level on Telecommunication


Usage Using Big Data
J. Palimote1, C. Aloy-Okwelle2, O. T. Olise 3
Department of Computer Science, Rivers State University, Port Harcourt, Nigeria1,2,3

Abstract:
Telecom network often encounters large number of tweets based on the user experience for a network. This
huge amount of raw data can be used for industrial or business purpose by organizing them according to our
requirement and processing. The aim of this paper is to address the social media review challenges in telecom
companies. The methods include; extracting tweets, analyzing them and segregating them into various
categories to help the company understand the concerns of their customer. This can help save millions and
prevent customer churn. In other to build a robust model, the dataset was pre-processed by checking and
removing Nan values. After the pre-processing, stage the cleaned data was tokenized. In tokenization process,
each word was divided into tokens, for easy training. After the tokenization process, we performed an
exploratory data analysis on the dataset to understand the pattern of the dataset. After the explorative data
analysis stage, we trained a random forest classifier to predict/classify the customer’s satisfaction into positive,
negative, and neural.

Keywords: Big Data, Telecommunication, Machine Learning, Random Forest classifier


1. Introduction
Big data can be used to analyze customer satisfaction levels in the telecommunications industry by collecting
and analyzing large amounts of data from various sources such as customer surveys, call center logs, and social
media. This data can be used to identify patterns and trends in customer usage, preferences, and satisfaction.
Machine learning algorithms can also be applied to this data to predict customer behavior and identify potential
issues or areas for improvement. Big data can be used to track and monitor the performance of
telecommunications networks and identify potential bottlenecks or areas for optimization. In order to analyze
customer satisfaction levels on telecommunication usage using big data, its essential to gather a large
amount of data on customer usage and behavior. This can include data on calls made, text messages sent,
internet usage, and any other relevant information. Once this data has been collected, various big data tools
and techniques, such as machine learning, is used to analyze and understand patterns and trends in the data.
This can help identify areas where customers are experiencing issues or dissatisfaction, as well as areas where
the service is performing well. Additionally, this data can be used to develop and implement strategies for
improving customer satisfaction and addressing any issues that are identified.

2. Related Work
[5] Provided a methodology for telecom companies to target different-value customers by appropriate offers
and services. This methodology was implemented and tested using a dataset that contains about 127 million
records for training and testing supplied by Syriatel Corporation. Firstly, customers were segmented based
on the new approach (Time- frequency- monetary) TFM (TFM where: Time (T): total of calls duration and

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25801


Internet sessions in a certain period of time. Frequency (F): use services frequently within a certain period.
Monetary (M): The money spent during a certain period.) and the level of loyalty was defined for each
segment or group. Secondly, the loyalty level descriptors were taken as categories, choosing the best
behavioral features for customers, their demographic information such as age, gender, and the services they
share. Thirdly, several classification algorithms were applied based on the descriptor and the chosen features to
build different predictive models that were used to classify new users by loyalty. Finally, those models were
evaluated based on several criteria and derive the rules of loyalty prediction. After that by analyzing these
rules, the loyalty reasons at each level were discovered to target them the most appropriate offers and services.
[4] Developed their own predictive models. The “Logistic regression model Design”. The model used
customer data to predict customer retention in a telecommunications company. The model predicts customer
retention based on billing, value-added services, and SMS service issues with. The system has a 95.5%
accuracy.
[1] Focused on using machine learning to determine the value of customers in the hospitality sectors. The
engaged customers by introducing the dynamic loyalty program. Their results show
that automated learning processes performed well in identifying the customers with greater value in specific
promotions. There work contributed in deepening the practical and theoretical understanding of automated
learning in the value chain of customer loyalty, in a structure that uses a dynamic model for customer
engagement.
[2] Provides a retrospect on how telecom operators have been striving, before the era of big data,
in analyzing large volumes of data in business operation. Their work examined the driving forces of big data
analytics in the telecom domain. They explored the benefits and the associated challenges.
[4] Explored ways of integrating big data insights with automated and assisted processes related to key
customer touch points to ultimately improve the customer experience. They provided an insight on how the
Alcatel-Lucent and Bell Labs helps the CSPs improve their business performance, using a unique
methodology. This can be used to improve the Net Promoter Score (NPS) and a higher customer value.

3. DESIGN METHODOLOGY

This session discusses the technique used and the methods used in the data collection. The system
architecture of the proposed system can be seen in Figure 1.

Preprocess
Tele-Communication Data ing
Tokenization

Dataset Data Filtering Lemmatizati Stemming


Data Cleaning By Length on

Model Evaluation ExploratoryData


Analysis

Figure 1. Architecture design of the proposed system

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25802


Dataset (Tweets): The dataset contains thousands of tweets from different customers concerning their
satisfactions towards Telecommunication Company. The dataset is an unstructured one. It was collected via
web scrapping using beautiful soup technique.

Pre-processing: This involves the processing of the legal case document, so that it would be fit for a training
a deep learning model. The following are the process of tokenization:

i. Data Cleaning and Noise Removal: One of the key steps in processing language data is to remove
noise so that the machine can easily detect the patterns in the data. Text data contains a lot of noise, in
the form of hashtags, punctuation and numbers.
ii. Filtering by length: It is useful to remove unwanted words in a sentence. Words that are usually less
than two characters in length do not represent a special meaning in a sentence. However, these words
have characteristics that are not defeated in the previous pretreatment process. Therefore, in this
process, only the necessary words are split by limiting the length of the words
iii. Transform cases: In this process, each character in a word is converted to lowercase.

Tokenization: Tokenization is the process of splitting the text into smaller pieces called tokens. Words,
numbers and punctuation marks and others can be considered as tokens. Each sentence in the document was
separated by words. To achieve this, the nltk (Natural Language Tool Kit) package in Python programming
language was used.

i. Stemming and Lemmatization: The aim of stemming is to inflectional forms to a common


base form. For grammatical reasons, text documents are going to use different forms of a word such as
take, taken, taking.

Exploratory Data Analysis (EDA): This was used in performing visualization of the data. This was used in
analyzing the data using charts and graphs.
Model Evaluation: The Random Forest Regression algorithm is used in order that to find the output from the
training dataset, it uses multiple decision trees. The mathematical representation of the random forest can be
seen in Table 1.

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25803


Table 1. Mathematical Representation of Random Forest Model

Impurity Task Formula Description


Gini Impurity Classifications 𝐶 𝑓𝑖 is the frequency of
∑ 𝑓��(1 − ����)
��=1 label I at a node and C is
the number of unique
labels
Entropy Classifications 𝐶 𝑓𝑖 is the frequency of
∑ −����log(��) label I at a node and C is
��=1
the number of unique
labels
Variance/Mean Regressions 1 𝑁 𝑦𝑖 is the label for an
∑ (𝑦𝑖 − ��) 2
Square Error 𝑁 ��= instance, N is the
number of instances and
1 μ is the mean
Variance/Mean Regressions 1 𝑁 𝑦𝑖 is the label for an
∑ |𝑦𝑖 − ��|
Absolute Error 𝑁 ��= instance, N is the
number of instances and
1 μ is the mean

4. Implementation/Result
This system presents a model for analyzing customers’ behavior on telecommunication using big data. The
system starts by acquiring unstructured data, to address the social media review challenges for
telecommunication companies and then analyzing the data, and segregating them into various categories to
help the company understand the concerns of their customers. In other to build a robust model, the tweet
dataset was pre-processed by checking and removing NaN values. The cleaned data can be seen in Figure 2.
After the stage of pre-processing, the cleaned data was tokenized. By tokenization, we divided each word into
tokens, for easy training and understanding, this is seen in Figure 2. After the tokenization process, we
performed an exploratory data analysis on the dataset in order to understand the dataset’s pattern. This is
displayed in Figures 3, 4, and 5. After the explorative data analysis, we trained a random forest classifier to
predict/classify the customer’s satisfaction into positive, negative, and neural. The result of the random Forest
model can be found in Figures 6 and 7.

Figure 2. Cleaned Data

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25804


Figure 3. Histogram of the customer’s satisfactory

Figure 4. Pie Chart of Customer’s reaction towards Tele communication

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25805


Figure 5. Word Count

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25806


Figure 6: Histogram of the total comments made by customers

Figure 7. Predicted Features of the Random Forest

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25807


5. Conclusion and Recommendation
This paper presents a model for the prediction and analysis of customer satisfaction in telecommunication. In
this paper, we examined a large amount of Verizon page tweets extracted from Twitter for a period of 4 days
using the hashtag #Verizon, classifying the tweets into different labels based on the user experience for the
network. We also analyzed the tweets based on their categories. Visualization of the dataset was carried out to
give a graphical representation of the user experience. The random forest algorithm was used for
classification. We can further recommend the following.
1. From our analysis we can say that Verizon must focus on their service (which includes customer service and
Network issues). By focusing on these issues they can improve their customer satisfaction and avoid churn
(which is also visible in the plots)
2. Using real time tweets to analyze the problem with the Verizon network and can provide a quick
solution to the problem and avoid network traffic. This can provide quality service to the customer.
3. Location based data can be used to offer good deals to customers and design campaigns to combat
churn.

Reference

1. Aluri A, Price BS, McIntyre N. (2019). "Using machine learning to concrete value through dynamic
customer engagement in a brand loyalty program". J Hosp Tour Res. 2019;43 (1):78–100.
2. Chung-Min Chen(2016) "Use cases and challenges in telecom big data analytics"Published online by
Cambridge University Press.APSIPA Transactions on Signal and Information Processing , Volume 5 ,
2016 , e19.DOI: https://doi.org/10.1017/ATSIP.2016.20
3. Jeffrey S., Yves T'J., Raluca D., Peter S., Laurent P. (2014). "Using big data to improve customer
experience and business performance" Bell labs technical journal 18 (4), 3-17,
2014
4. Oladapo K, Omotosho O, Adeduro O. (2018). Predictive analytics for increased loyalty and
customer retention in telecommunication industry. Int J Comput Appl. 2018;975:8887
5. Wissam N, Ramez A,, Kamal S, & Shadi B.(2020)"Predictive analytics using big data for
increased customer loyalty: Syriatel Telecom Company case study" J Big Data (2020)
7:29 https://doi.org/10.1186/s40537-020-00290-0 pp 2-24

J. Palimote, IJECS Volume 12 Issue 09 September, 2023 Page 25808

View publication stats

You might also like