Professional Documents
Culture Documents
1.0 Introduction.....................................................................................................................2
1.1 Problem Statement ..........................................................................................................5
1.2 Research Question ..........................................................................................................5
1.3 Research Objective .........................................................................................................5
2.0 Methodology ...................................................................................................................6
2.1 Tool .................................................................................................................................6
2.2 Data .................................................................................................................................6
2.3 Framework ......................................................................................................................7
3.0 Finding ..........................................................................................................................12
3.1 Data Collection .............................................................................................................12
3.2 Amazon .........................................................................................................................12
3.3 eBay ..............................................................................................................................15
4.0 Discussion .....................................................................................................................18
5.0 Conclusions...................................................................................................................25
5.1 Limitation......................................................................................................................26
5.2 Recommendations.........................................................................................................26
References................................................................................................................................28
Appendices...............................................................................................................................30
1
1.0 Introduction
It is known that there is abundant data generated every day all over the world, whether it is
structured or unstructured. One of the biggest contributors is online shopping.
The world spends around 1 million dollars per minute on commodities on the internet
(SeedScientific.com). Online shopping has become the norm with every passing day, and every
online journey of buyers is well documented typically, which benefits each purchase and can
contribute how much data is created every day. Moreover, the online community platforms
also are one of the biggest ones to contribute data such as Twitter, Facebook, and Snapchat.
often post their anecdotes or comments that are satisfied, angry or disappointed with certain
products they buy on online communication platforms such as Twitter. Thus, there are more
service. Other users also can repost this post on his or her account so that it makes more and
more people know this post, which can increase the number of followers of the person who did
the post. It can attract more people to comment on their own opinions.
Nowadays, Twitter has already become one of the most famous social networking platforms.
It means that Twitter will collect a lot of data every day. According to the research in May
2
2020, around 6,000 tweets will be posted on average every second. In other words, 500 million
tweets will be sent each day (David Sayce.com). In this case, we will utilize the tweets of
Amazo
more and specific real-time updating data from varieties of users.
Sentiment Analysis
3
business sites, such as for products, services, politics, and movies. Sentiment Analysis helps
people to classify comments and opinions as polar views that are positive, negative, and neutral
based on the scores of given contents.
Sentiment Analysis is a study of text, which is widely used on comments and surveys through
a campaign according to the feedback or responses on commercial sites. In this case, the
sentiment analysis can help both
which may result in improvement of the sales and popularity as it tells the choices of most
customers or citizens.
Sentiment analysis has a high efficiency using machine learning approach, Lexicon-based
approach and Hybrid Techniques approach to extract and define sentiment content in a text
unit. In this case, Vader will be applied in sentiment analysis. It is a model which is used for
test sentiment analysis that is based on lexicons of sentiment-related words. In this approach,
be rated as negative. Hence, the Vader approach will help this study more conveniently and
faster to get the result of sentiment analysis.
eBay is an online auction and trading company launched in 1995, one of the first companies to
create an internet market web site to make buyers and sellers trade goods and services. It means
that eBay does not have its own products. It is an American multinational e-
commerce corporation based in San Jose, California, that facilitates consumer-to-
consumer and business-to-consumer sales through its website. Amazon is an online retailer,
manufacturer of electronic book readers and web services provider. In other words, except
third-party sellers, it includes its own brand product. (Britannica.com). They are both set in
America, two of the most famous online shopping platforms and are rivals for each other.
According to the research, there are 182 million eBay users and 150.6 million mobile users
4
worldwide in 2019 (OBERLO.com). It indicates that there are many purchases and orders will
be placed in one day, which will generate a huge amount of data. Not only that, many tweets
also about #Amaon and #eBay appear on Twitter with the development of eBay and Amazon.
With the time goes by, the people are attracted to use eBay or Amazon will be more and more,
but different people have their own preferred online shopping platform in their own minds.
As a matter of fact, there are business that embrace the failure in their brands reputation. This
phenomenon occurs due to lack of essential awareness of potential market, competitor, and
customers. They had totally neglected the impact of social media to their brands. Knowing and
understand the customer is important in maintaining a positive brand perception. Sentiment
analysis is important in guiding a business in understanding the customer as well as the
competitor. They can be used to predict market behaviour and, replace time-consuming and
expensive conventional methods such as focus groups and surveys. With the use of sentiment
analysis, companies have the possibility to evaluate the social media health of a brand and to
compare it to the business's competitor's brand which allows to capture trends and positive or
negative tendencies (Zitnik, 2012).
5
1.4 Significant Research
The main purpose of this study is to explore the sentiment of Amazon and eBay, for instance
to obtained useful information and insight from the analysis. This information is important for
e-commerce to
positively than the brands of competitors. This sentiment analysis also can be used to focus on
the customer feedback that is negative to understand the improvement needed. Besides,
positive feedback can be used to understand the reason customer satisfy with the brand.
2.0 Methodology
2.1 Tool
The tool we are going to utilize is RapidMiner, a software platform that provides varieties of
data analysis functions such as data preparation, machine learning, text mining, sentiment
analysis, and so on. In this study, we will use RapidMiner for data collection, preparation, and
sentiment analysis.
2.2 Data
In this study, data will be collected by extracting from Twitter by RapidMiner, the data time
ranges are 05 May to 12 May (Amazon) and 04 May to 12 May (eBay).
6
2.3 Framework
1. Data collection
2. Data Preparation
3. Sentiment Analysis
4. Wordlist Generation
5. Visualization
Data Collection
RapidMiner includes a Search Twitter operator, which is to search for Twitter statuses. It can
specify a query and get this data query from Twitter update status, which includes additional
data with context of the status. In expert mode, it can specify additional search restrictions such
as result type, data limitation, and so on (RapidMiner.com). This process needs to be connected
#eBay
#Amazon as with English language selection. The limited data query is 757 and
800 on Amazon and eBay, respectively. Then, the extracted data will be saved in an excel file
into our PC and RapidMiner.
7
The data collection will be organized in the form of a table, and the whole attributes include
ID, Created-At, From-User, From-User-Id, To-User, To-User-Id, Language, Source, Text,
Geo-Location-Latitude, Geo-Location-Longitude, and Retweet-Count.
Data Preparation
Further on, before proceeding the sentiment analysis, additional process will be applied, which
enables us to observe data clearly. The process is as below Figure. 3.
8
Data preparation consists of selecting attributes, subprocess, Trimming, Removing Duplicates
and Replacing Missing Value. For attributes we select are Create-At, From-User and Text. The
subprocess includes five Replace operators as below Figure. 4, which is to replace unnecessary
parameters, and the entire process is:
2. Replace colon =:
3. Replace hashtag = #
4. Replace retweets = RT
After that, trimming the white spaces is important as there are some unnecessary white spaces
after retweet replacing. Next is removing duplicates to ensure accuracy of results. And then is
Sentiment Analysis
9
Figure 5. Sentiment Analysis Figure 6. Function expression
Wordlist Generation
Wordlist generation is a text processing consisting of 5 steps that are shown below Figure. 7:
Tokenization, Transforming Cases, Filter Stopwords, Stemming and Filter Tokens.
1. Tokenize
It is an operator for dividing the sentence in the document into separate words, which
is to split words from the text, so that the words can be used into the next sub process
(T. Verma, R. Renu, D. Gaur).
2. Transform Cases
It is an operator that converts all texts from upper cases into lower cases, which is to
avoid the confusion of words.
3. Filter Stopwords
It is to remove recurrent and unnecessary words, as stop words include prepositions,
pronouns, determiners, conjunctions, and so on. They will take up many spaces in
databases but do not have meaningful information.
10
4. Stemming
It is a process that changes suffi
5. Filter Tokens
It is a process that eliminates the tokens which are shorter than 5 characters or longer
than 15.
Visualisation
After completion of sentiment analysis, the suitable plot type needs to be selected to make the
result of data visualized. In this case, bar chart will be used to describe the quantity of positive,
neutral, or negative, pie chart will be used to indicate the probability they cover, respectively.
Wordcloud will be utilized to generate the most frequent words regarding the scoring string.
Finally, wordlist will be released out for most frequent words viewing.
11
3.0 Finding
3.2 Amazon
After completing the sentiment analysis, the sentiment overview for Amazon is as below:
Amazon
Polarity Count (Sentiment) Percentage (%)
Positive 458 77
Neutral 63 11
Negative 73 12
400
neutral
300
11%
200
positive
100 77%
0
positive neutral negative
As shown, the majority tweet for Amazon is positive with 458 tweets (77%) in this category.
A negative opinion was found to be 73 tweets (12%) and neutral opinion to be 63 tweets (11%).
The positive sentiment for amazon is far higher than neutral and negative sentiment.
12
Positive Sentiment
#Amazon 50 Top Frequent Word and Scoring String for Positive Sentiment
13
Negative Sentiment
#Amazon 50 Top Frequent Word and Scoring String for Negative Sentiment
14
3.3 eBay
After completing the sentiment analysis, the sentiment overview for eBay is as below:
eBay
Neutral 301 46
Negative 59 9
The majority opinion for eBay is considered neutral with 301 tweets (46%). Positive and
negative opinion was found to be 295 tweets (45%) and 59 tweets (9%) respectively. Positive
and neutral sentiment are relatively close for eBay while the negative sentiment is relatively
low.
15
Positive Sentiment
#eBay 50 Top Frequent Word and Scoring String for Positive Sentiment
16
Negative Sentiment
#eBay 50 Top Frequent Word and Scoring String for Negative Sentiment
The frequent word list for eBay negative sentiment is less difference as the positive sentiment
frequent words)
17
4.0 Discussions
through natural language processing (NLP). Sentiment analysis can be used to identify and
summarize customer opinion from user feedback which further improve the customer
experience and make it more favourable for the user to shop at the e-commerce platform. Both
positive and negative feedback help the user and the manufacturer or seller. Manufacturer or
seller can take the negative opinion constructively and can know the area that needed
improvement, and hence enhance their product or service. Amazon and eBay are chosen for
this project as they are direct competitor of each other despite of their difference platform
characteristic.
18
Figure 10. eBay e-Commerce Platform
Amazon having more positive sentiment (77%) as compared to eBay (45%). Both platforms
have low negative sentiment. This indicate that they are at a good performance which satisfy
majority of their user.
According to WebFx (2021), Amazon scores well for trust, loyal customers, lower fees,
branding opportunities and their fulfilment service. On the other hand, eBay is usually better
for competitiveness, lower fees, loyalty, and fewer restrictions. Amazon has initiate Lightning
Deals, Special Coupons and Prime Deals to 3rd party seller which enable seller to set a discount
for their product. Amazon giving the chance for seller to earn more business by creating sales,
promotion, and coupon code, which eBay lack of. Amazon is seeming to compete with eBay
with facilitating selling and strengthening its affiliates program. eBay had a different approach
on its retailing while Amazon were interested in opening online store where they can sell
products to consumers. Report stated that auction format alone limits the growth of eBay, hence
it started to pursue fixed-price retailing (Krishnamurthy, 2004).
19
Positive Sentiment
Noted the frequent wordlist for positive polarity is rather different for Amazon and eBay due
to the website characteristic. eBay is auction market where buyers and sellers enter competitive
bids simultaneously. Hence, finding shows that eBay are lack of marketing promotion word
come
from the promotion. Customer having positive feedback on the Amazon promotion. This prove
that marketing promotion is one of factor influencing customer satisfaction.
Elan Musk and several celebrities tweet regarding Dogecoin sent this cryptocurrency to a high
record (Browne, 2021). A change.org has also open up a petition previously requesting
Amazon to accept Dogecoin. Recently, Amazon has announced that they will not accept
Dogecoin as a payment method, however third-party applications can purchase Amazon gift
-issued currency. The frequent
word list that appears in amazon can be explained by the properties of the company itself that
served as provider of consumer electronics, online and non-online retail services, and other
products.
Some positive feedback for Amazon adopted from the original tweets include:
This is very amazing and great step by thankyou amazing (0.72) 3.20
i used amazon in uae last 4 year its great and great (0.79)
amazing product amazon (0.18)
great (0.79)
amazing (0.72)
Just saw this on Amazon Michael Kors Jet Set amazon (0.18) 0.51
Travel Large Trifold Leather Wallet (Admiral admiral (0.33)
Silver Hardware) by Michael Kors for $69.99
20
If you've been a Kindle Unlimited subscriber in amazon (0.18) 0.18
the past, Amazon are offering six months for half-
price for returning customers.
Since eBay mainly focus on auction, the frequent word occur are quite related to this activity
competitive as compared to Amazon as they do not sell their own product while Amazon have
their own brand. However, both platforms have a massive reach of audience but with difference
eBay. Nintendo is a Japanese multinational consumer electronics and video game company.
The famous game for this company includes Mario, Pokemon and animal crossing. The
frequent occurrence of this word suggests that the demand for the video games is relatively
high. According to The DealExpert (2020), the Nintendo Switch have no supply available in
the US, hence the cost of the Nintendo Switch on the remaining supply tends to be marked up,
by third-party sellers or resellers. In eBay there are more Nintendo Switch for sale, and mostly
are sold at retail prices.
eBay deals with a wide range of customers and products. Auction facilities, instant buying,
bidding, and classified services are the main service this platform offer. It is a platform where
user can find antiques, collectibles, new and second-hand stuff. Majority of eBay deals are in
their auction section. It is a place where seller can put their unique products on auction. Buyer
from eBay are usually collectors of unique and specific product which is antique and
collectibles (Chaffey, 2021). This explained the frequent word list in the sentiment analysis
eBay which is rather different than that occur in Amazon.
21
Some positive feedback for eBay adopted from the original tweets include:
Negative Sentiment
Both Amazon and eBay show relatively low negative sentiment, where negative sentiment are
from minority of the failure, cancelled promotion and shipping.
Some negative tweets from Amazon adopted from the original tweets include:
Why Amazon is number one and they will amazon (0.18) -0.90
remain ? They prefer loose money instead of
number (0.08)
doing what you are doing to their customers.
(Tickets opened weeks with no feedback and no loose (-0.33)
22
But my order was canceled and I never fuck (-0.64) -0.6
received anything. Now is charging my card
amazon (0.18)
again bc I never returned something. How the
fuck so you return something that was canceled
& you never received?! Amazon
eBay however have built a reputation for hosting scammers in some aspect. This have been
improved after eBay establish term to protect their buyer if the items do not arrive. Amazon is
stricter on their seller regulation where harsh punishment will be given to their seller if they
are found out to be foul of Amazon rules and regulations (Dzieza, 2018).
Some negative tweets from eBay adopted from the original tweets include:
paypal is making scam eBay should change its scam (-0.69) -0.69
payment method!
23
Suggestions
As a suggestion, Amazon and eBay can reward their customer for taking action like sharing
and hastagging product, experience and review. By doing so, this e-Commerce would have
more user data, increase engagement, and reap the benefit in term of purchase rate and order
review. Data have shown that most of the time, customer gain more information from review
and opinion as compared to the description on the website itself. The opinion of users are said
to influence the decision of another buyer (Helverson, Abramczuk, Kopec & Nielek, 2018).
Besides, business could offer campaign which could attract more customer such as year end
sales and promotion. It is also important for business to stay ahead of response as social media
is quick and customer are using them to discuss their experience positive or negative and it is
visible to their whole following. Customers are usually expecting response withing hours, so it
would be helpful for brands to connect it them by responding to all types of feedback quickly
on whatever platform the customer are using.
24
5.0 Conclusions
With the increasing of the market competition, sentiment analysis become an important process
to understand what consumers think about a product or brand. Sentiment analysis can assist
organization to understand what their users think and feel about their brand. Review or opinion
can be collected from difference source such as social media platform, review sites, apps stores
and eCommerce stores. One of the tools for sentiment analysis is Rapid Miner which is capable
of mining online opinion for sentiment analysis. Vader is a model used for text sentiment
analysis which is sensitive for polarity and intensity of emotion. It relies on a dictionary that
maps lexical features to emotion intensities known as sentiment scores. The scores are finalized
and used to classify the opinion into positive, negative, or neutral opinions. However, sometime
labelling word as positive or negative alone is not enough for any decision making, hence it is
important to investigate the reason behind the polarity of the opinion (Poonguzhali, Waldiya,
Vinothini & Livisha, 2018).
From the data obtained, the majority sentiment for Amazon is positive while the main sentiment
for eBay is neutral. Though the difference between Amazon and eBay is in Amazon is a retailer
selling product to consumer while eBay act as an auction house where consumer congregate to
sell to one another, they are companies of direct competitor. Study reviews that eBay tend to
receive more visitor as compared to Amazon back in 2002 holiday season. eBay has grown
approximately 167 million buyer and around 25 million sellers (The Strategy Watch, 2021).
25
5.1 Limitation
5.2 Recommendations
For further analysis, sentiment analysis can be done on the official webpage for a more
comprehensive data collection. More data can be collected at different time being. Besides,
clustering process might need towards the tweets gathered to identify more suitable dimension
to represent the online platform quality that influence the customer feedback. Furthermore,
customer rating can be included to provide more useful information to improve customer
experience on the platform. Sentiment analysis can be great benefit to the industry and require
researcher to gather less time consuming, cutting down the expenses incurred using surveys,
questionnaires, interviews, market research and trends.
Competitive Analysis Sentiment analysis in business allows the business to find gaps in
their marketing strategy, manage their brand reputation, and zero
in on key areas where customer sentiments are positive or negative.
Brand Perception Knowing the customer and understanding them is very important
for building and maintaining a positive brand perception.
Businesses can grow only when they truly understand the people
26
using their products or services. Besides, quick action can be take
one negative sentiment is detected before they turned into disaster.
(Bianchi, 2021)
27
References
Browne, R. (2021). CNBC. Tweets from Elon Musk and other celebrities send dogecoin to a
record high. Retrieved from https://www.cnbc.com/2021/02/08/tweets-from-elon-
musk-and-celebrities-send-dogecoin-to-a-record-high.html
Chaffey, D. (2021). Smart Insight. eBay marketing strategy case study. Retrieved from
https://www.smartinsights.com/ecommerce/ecommerce-strategy/ebay-case-study-2/
Hall, M. (n.d.). Amazon.com | History & Facts. Encyclopedia Britannica. Retrieved May
30,2021, from https://www.britannica.com/topic/Amazoncom.
Helversen, B., Ambramczuk, K., Kopec, W., & Nielek., R. (2018). Influence of consumer
reviews on online purchasing decision in older and young adults. Decision Support
System. 113(1-10).
Poonguzhali, R., Vishal Waldiya, S., & Vinothini, K. L. (2018). Sentiment Analysis on
LinkedIn Comments, International Journal Of Engineering Research & Technology
(Ijert) Iconnect, 6(7).
Sayce, D. and Ltd., P., 2021. The Number of tweets per day in 2020 | David Sayce. [online]
David Sayce. Available at: https://www.dsayce.com/social-media/tweets-day/
28
Verma, T., Renu, R., Gaur, D.,
International Journal of Applied Information Systems, vol. 7, pp. 16-18, 2014.
Vuleta, B., 2021. How Much Data Is Created Every Day? [27 Powerful Stats]. [online]
SeedScientific. Available at: <https://seedscientific.com/how-much-data-is-created-
every-day/>
WebFX. (2021). Amazon vs eBay for business: Which one is better? Retrieved from
https://www.webfx.com/amazon/ebay-or-amazon-for-businesses.html
29
Appendices
30
Appendix II Top 50 Frequent Word in Amazon Negative Sentiment
31
Appendix III Top 50 Frequent Word in eBay Positive Sentiment
32
Appendix IV Top 50 Frequent Word in eBay Negative Sentiment
33