Social Media Analysis for Understanding Customer Preferences and Sentiments (BUS5CA) By Student Name> (18752031) [pic] La Trobe Business School Melbourne, Australia

Customer Analytics with Social Media
Assignment 1: Social Media Analysis for Understanding Customer
Preferences and Sentiments
(BUS5CA)
By
<Student Name>
(18752031)
La Trobe Business School
Melbourne, Australia

Table of Contents
1. Introduction 1
2. First Case Study 1
2.1 Data Exploration: Article properties 1
2.1.1 Statistics of Shares 1
2.1.2 Characteristics of high number of shares 7
2.2 Keyword analysis using SAS Text Miner 11
3. Second Case Study 15
References 17
Appendix A

List of Figures
Figure 1 Overall Shares vs Title length...........................................................................................8
Figure 2 Overall shares vs Content length......................................................................................9
Figure 3 Overall shares vs Published in weekends..........................................................................9
Figure 4 Topics of whole data vs Worth.......................................................................................15
Figure 5 qplot of Emotions............................................................................................................16

List of Tables
Table 1 Data channel “Lifestyle”: Top 10 Shares 2
Table 2 “Data channel “Entertainment”: Top 10 Shares 2
Table 3 “Data channel “Business”: Top 10 Shares 3
Table 4 “Data channel “social media”: Top 10 Shares 4
Table 5 “Data channel “technology”: Top 10 Shares 5
Table 6 “Data channel “world”: Top 10 Shares 6
Table 7 Channel “Lifestyle”: Top 10 topics 11
Table 8 Channel “Entertainment” : Top 10 topics 12
Table 9 Channel “Business”: Top 10 topics 12
Table 10 Channel “Social Media”: Top 10 topics 13
Table 11 Channel “Technology: Top 10 topics 13
Table 12 channel “World”: Top 10 topics 14
Table 13 “Complete” data: Top 10 topics 14
Table 14 Eight Emotions Counts 16

1. Introduction
Text mining and sentiment analysis, helps to covert unstructured data to meaningful insights, and can be
explained as computation of polarity of emotions and views, and is most popular method for language
processing (Nikhil et al., 2015). For the case study one, our main aims are the detection of the most
popular articles shared on the social media and their characteristics use text mining method. This analysis
would help to leveragethe power of content and using planned social media marketing, this can assist in
sharp rise in customer base and audience. For case study 2, we will mainly focus to extract
emotions/sentiments from the movie review tweets using dictionary based sentiment analytics method.
For sentiment analysis, the process includes different stages like identification of polarity (i.e. positive,
negative, or neutral) of text and extracting the emotions or sentiments (Ghiassi et al., 2016). This analysis
would provide an insight to build better brand awareness. And marketing strategy by gauging attitudes of
customers.
2. First Case Study
1.1 Data Exploration: Article properties
Under life style channel, number of record is 2100, under entertainment channel, number of record is
7059, under business channel, number of record is 6259, under social media channel, number of record is
2325, under technology channel, number of record is 7345, and under world channel, number of record is
8425. But, no associations with channels were observed for 6147 records.
1.1.1 Statistics of Shares
Highest number of shares i.e. 843300 were observed for the article “Leaked: More Low-Cost iPhone
Photos”, this article not associated with previously mentioned six channels.
1

Lifestyle
Under lifestyle channel, highest number of shares i.e. 208300 were observed for the article ‘Obama to
Discuss NSA Reform with Lawmakers’. Table 1 contains top ten shares under lifestyle channel.
Table 1 Data channel “Lifestyle”: Top 10 Shares
Article Title Number of
shares
‘Obama to Discuss NSA Reform With Lawmakers’ 208300
‘No Movie Trailer Is Complete Without This One Line’ 196700
‘87% of American Teenagers Send Text Messages Each Month’ 139600
‘High-Tech Wristband Monitors Mood’ 81200
‘22 Books for Your Ultimate Summer Reading List’ 73100
‘Finalists Exhibit Tech for $465 Million Virtual Border Fence’ 56000
‘Cybersecurity Experts Will Face Off in Mock NetWars’ 54900
‘84% of Smartphone Owners Use Apps While Getting Ready in the
Morning’
54200
‘It's Still Easy to Get Away With Revenge Porn’ 49700
‘Beats SoloA Headphones Sound Great, But You're Paying for Fashion
[REVIEW]’
45100
Entertainment
Under entertainment channel, highest number of shares i.e. 210300 were observed for the article ‘Sprint's
New Plans Guarantee Unlimited Data for Life’. Table 2 contains top ten shares under entertainment
channel.
2