Social Media Analysis for Understanding Customer Preferences and Sentiments

Added on -2020-02-19

| BUS5CA| 43 pages| 4170 words| 63 views

Trusted by 2+ million users,
1000+ happy students everyday

Showing pages 1 to 8 of 43 pages

Customer Analytics with Social Media
(Assignment 1 - Social Media Analysis for Understanding Customer
Preferences and Sentiments)
By
<Student Name>
(18815197)
La Trobe Business School
Melbourne, Australia
Table of Contents
1. Introduction 1
2. Case Study 1 2
2.1 Explore the impact of article properties 2
2.1.1 Channel wise number of shares 2
2.1.2 Properties of high number of shares 9
2.2 SAS Text Miner for keyword analysis 14
3. Case Study 2 18
References 20
Annexure i
List of Figures
Figure 1 Number of shares vs Title length (Overall).................................................................................10
Figure 2 Number of shares vs Content length (Overall)............................................................................10
Figure 3 Total shares vs Published on the weekend (Overall)...................................................................11
Figure 4 Comparison of average title lengths of top 10 shares of each of the channel...............................13
Figure 5 Comparison of average content lengths of top 10 shares of each of the channel.........................13
Figure 6 Comparison of no. of publishes in weekdays of top 10 shares of each of the channel.................13
Figure 7 Topics of whole data vs Share.....................................................................................................17
Figure 8 qplot of sentiments......................................................................................................................19
List of Tables
Table 1 Top 10 shares under “Lifestyle” channel 3
Table 2 Top 10 shares under “Entertainment” channel 4
Table 3 Top 10 shares under “Business” channel 5
Table 4 Top 10 shares under “Social media” channel 6
Table 5 Top 10 shares under “Technology” channel 7
Table 6 Top 10 shares under “World” channel 8
Table 7 Top 10 topics under “Lifestyle” channel 14
Table 8 Top 10 topics under “Entertainment” channel 14
Table 9 Top 10 topics under “Business” channel 15
Table 10 Top 10 topics under “Social Media” channel 15
Table 11 Top 10 topics under “Technology” channel 16
Table 12 Top 10 topics under “World” channel 16
Table 13 Top 10 topics under “Whole” data 16
Table 14 Top 10 topics under “six channels”: Cumulative 17
Table 15 Eight Emotions vs Counts 18
1. Introduction
Emotions are an integral ingredient of human life. Through the visible gestures, it can directly affect the
audience. It is very difficult to comprehend someone’s intention or sentiments only through texts without
any visibility. With the rise of digital era, and with the wide use of mobile, internet, digital media and
social networks, textual communication has turned out to be the main source of communication. Hence,
need has been felt in research community (Pandey et al., 2016) to create a model which can recognize the
sentiments behind the text and can make these digital communications more effective (Li and Wu., 2010;
Pang and Lee, 2008). This study focuses on text mining and sentiment analysis, which are nowadays most
popular methods for language processing. Polarity of emotions and views can be computed using these
models, these models assist to create meaningful insights out of unstructured data. In this study, our main
objectives are the identification of the impact of articles on sharing based on the metadata and to
understand sentiment of movie reviewers in twitter and comparison of different methods of sentimental
analysis. In case study 1, our objective is to find out the relationship between different keywords, shares,
and sentiments, which would help to make strategy for improved online advertising and better
communication with customers. In case study 2, our objective is to predict consumers’ review sentiments,
which would help to provide meaningful insights to improve the campaign tracking, customer-oriented
marketing strategy and brand awareness. During sentimental analysis, our target will be polarity
identification of twitter comments (positive, neutral, or negative) and mining the characteristics of a given
comment to uncover their polarities (Pak and Paroubek, 2010).
1
2. Case Study 1
1.1 Explore the impact of article properties
As mentioned in the assignment, six excel sheets (lifestyle (lifestyle), entertainment, business (bus), social
media (socmed), technology (tech), and world (world)) have been formed from the provided data i.e.
“OnlineNewsPopularity.csv”. Few facts which have been identified while exploring the data, have been
mentioned below.
Observations for life style, entertainment, business, social media, technology, and world are 2100,
7059, 6259, 2325, 7345, and 8425 respectively. There are and 6147 observations which have no
associations with any of these channels. There are also few observations which are associated with two
channels like “Title 157”, which is associated with entertainment & social media both; title 24 is
associated with lifestyle & social media both; title 185 is associated with business & social media both;
and title 345 is associated with business & technology both.
1.1.1 Channel wise number of shares
“Leaked: More Low-Cost iPhone Photos” has got highest number of shares i.e. 843300, though this is not
falling under the six channels.
Lifestyle
For lifestyle channel, “Obama to Discuss NSA Reform with Lawmakers” has got highest number of
shares i.e. 208300. The top 10 shares under Lifestyle channel have been shown below in Table 1.
2
Table 1 Top 10 shares under “Lifestyle” channel
Sl.
No
.
url Title shares
1 http://mashable.com/2014/01/07/
obama-nsa-reform-lawmakers-
meeting/
Obama to Discuss NSA Reform With
Lawmakers
208300
2 http://mashable.com/2013/07/08/
supercut-one-man-trailers/
No Movie Trailer Is Complete Without This
One Line
196700
3 http://mashable.com/2013/11/25/
teenage-online-activity/
87% of American Teenagers Send Text
Messages Each Month
139600
4 http://mashable.com/2013/06/11/
wristband-mood-monitor/
High-Tech Wristband Monitors Mood 81200
5 http://mashable.com/2013/05/29/
summer-reading-list/
22 Books for Your Ultimate Summer
Reading List
73100
6 http://mashable.com/2013/07/11/tech-
virtual-border-fence/
Finalists Exhibit Tech for $465 Million
Virtual Border Fence
56000
7 http://mashable.com/2013/12/10/
mock-netwars/
Cybersecurity Experts Will Face Off in
Mock NetWars
54900
8 http://mashable.com/2013/10/15/apps-
morning-commute/
84% of Smartphone Owners Use Apps While
Getting Ready in the Morning
54200
9 http://mashable.com/2013/10/21/
revenge-porn/
It's Still Easy to Get Away With Revenge
Porn
49700
10 http://mashable.com/2014/05/29/beats-
solo-2-review/
Beats SoloA Headphones Sound Great, But
You're Paying for Fashion [REVIEW]
45100
3
Entertainment
Sprint's New Plans Guarantee Unlimited Data for Life” has received maximum number of shares
(210300) under entertainment channel. The top 10 shares under entertainment channel have been shown
below in Table 2.
Table 2 Top 10 shares under “Entertainment” channel
Sl.
No
.
url Title shares
1 http://mashable.com/2013/07/12/
sprint-unlimited-data-for-life/
Sprint's New Plans Guarantee Unlimited Data
for Life
210300
2 http://mashable.com/2013/12/25/
xbox-one-getting-started/
What to Do With Your New Xbox One 197600
3 http://mashable.com/2013/12/26/
mcdonalds-kills-mcresource-line/
McDonalds Kills Site That Advised Employees
to Eat Healthy Meals
193400
4 http://mashable.com/
2013/08/28/6000-video-launched-
helloflo/
How a $6,000 Video Got 6 Million Views and
Launched a Business
138700
5 http://mashable.com/2014/02/10/
flappy-bird-typing-tutor/
'Flappy Bird Typing Tutor' Is Even More
Frustrating Than the Original
112600
6 http://mashable.com/2014/10/14/
sandworm-russian-hackers-nato-
with-microsoft-bug-ukraine-nato/
Russian Hackers Used Microsoft Bug to Spy
on Ukraine and NATO
109500
7 http://mashable.com/2014/05/28/
lookout-theft-protection/
Lookout Fights Back Against Smartphone
Thieves
109100
8 http://mashable.com/2014/09/10/
australian-tested-ebola-virus/
Australian Patient Tests Negative for Ebola 98500
9 http://mashable.com/2014/01/14/
facebook-yandex-partnership/
Facebook Makes Inroads in Russia With
Yandex Partnership
98000
4

Found this document preview useful?

You are reading a preview
Upload your documents to download
or
Become a Desklib member to get accesss

Premium

$45

Q&A Library Access

Chat support

12

Document Unlocks

4

Answer Unlocks

Students who viewed this