BUSINESS ANALYTICS

ASSIGNEMENT

Big Data Application

in E-commerce:
Case – Amazon

DARSHAN
22BC404
SEM – 6
SECTION - G
 I. INTRODUCTION
In a variety of e-commerce applications, big data analytics has become a new interface for
competitiveness and innovation. It offers a huge number of businesses that turn data into
insights to make logical decisions and provide answers to company challenges by utilizing
people’s dynamics, processes, and technologies. To get a competitive edge, this strategy
necessitates the close integration of data, resources, skills, and systems. Prominent online
retailers like Google, Amazon, and Facebook have realized that big data analytics contributes
to their yearly income growth.

In the American Express Awards Customer Choice Awards for 2010 and 2011, respectively,
leading online retailer Amazon and its subsidiary Zappos were listed among the top ten
stores.

Industry watchers agree that Amazon’s use of big data resources to deliver higher-than-
average service quality was the reason for the company’s dominance in this field. Along with
many other significant Internet companies, Amazon started focusing on big data to enhance
its performance after emerging as the industry’s leading ISP in the early 2000s.
That has since shifted to making appropriate use of large databases pertaining to consumers
who were making purchases through e-commerce platforms. Zappos' acquisition in 2009
allowed Amazon to leverage big data to enhance customer service quality and validate
organizational fraud. Not only did it allow us to easily access customer profiles, purchasing
behavior, and browsing history, but it also allowed our executives to quickly respond to
customer inquiries.

Big data has been instrumental in shaping Amazon’s business strategy, enhancing customer
relationships, and driving market competitiveness. The analysis delves into Amazon’s
strategic approach, including the adoption of sophisticated data management systems—
ranging from real-time to batch processing, and handling both structured and unstructured
data. Additionally, Amazon’s insights into customer preferences and behaviors, highlight its
consequent ability to cater to diverse market segments. This comprehensive approach has
necessitated significant investments in evolving data storage models to meet high-speed
transaction demands. The findings aim to provide actionable insights for other e-commerce
entities looking to leverage big data effectively.
II. CHARACTERISTICS OF BIG DATA

1. Volume

Amazon, one of the world’s largest e-commerce and cloud computing companies, is one of
the largest traffic sources on the Internet today, accounting for about 3.1% of total
HTTP/HTTPS traffic, or one-fifth of Google’s. We deal with huge amounts of data every day,
and the data is very broad and complex, covering multiple business areas, including e-
commerce, cloud computing, artificial intelligence assistants (such as Alexa), digital media
(such as Amazon Prime Video and music), logistics and supply chain, etc., and we deal with
enormous scale and diversity of data every minute of every day including order data, user
browsing behavior, cloud service performance indicators, and more. Today, AWS powers
hundreds of thousands of businesses in 190 countries around the world.

2. Velocity

Amazon employs multiple technologies and strategies in different fields and applications to
enable fast and efficient data processing, with almost every book at your fingertips in 60
seconds. Amazon’s data is generated and updated in real time. This is evident in quickly
processing customer orders, tracking shipments, and real-time inventory man- agement.
Amazon uses the following methods to improve speed:

• Real-time processing (Apache Kafka and Amazon Kinesis)

• Batch technology (Apache Hadoop and Amazon EMR)
• Distributed computing (Apache Spark)
• High-performance storage (Amazon S3, Amazon RDS, and Amazon DynamoDB)
• Caching technology (Amazon ElastiCache)

3. Variety

Due to the company’s diversified business and global operations, Amazon involves multiple
business areas and data types. Amazon processes various types of data, including order
history, customer profiles, customer reviews, product descriptions, text reviews, images, and
more. They also process data in a variety of formats, utilizing a variety of data sources and
formats.

4. Veracity

The quality and accuracy of the data are of Paramount importance to Amazon. Amazon
invests heavily in data vali- dation, quality control, and data governance to minimize errors
and maintain the integrity of data. Amazon uses a variety of methods and strategies to ensure
data accuracy:

• Data validation and cleansing

• Data quality monitoring
• Data audit and tracking
 • Data security
• Data redundancy and backup

5. Variability

The needs and behaviors of customers change greatly, especially during peak shopping
season or due to external factors, the data is constantly changing. Amazon has responded to
these changes with a deep understanding of shop- ping behavior and increased investment in
machine learning algorithms and has achieved remarkable results. For example, Amazon can
tailor the placement of sponsored products to the user’s profile, making it relevant to what the
customer searches for, which makes the ads more useful to the customer.

6. Visualization

Amazon uses a variety of tools and techniques to visualize data to better understand and
interpret its vast amounts of data and support data-driven decisions.

• Amazon QuickSight is Amazon’s cloud data visualization service that allows users to
create interactive dash- boards and reports.
• Amazon SageMaker is Amazon’s machine learning service that can be used for data
visualization, especially during model training and deployment.

7. Value

At the heart of Amazon’s business model is extracting value from data. By analyzing
customer behavior, Amazon can provide personalized product recommendations, optimize
logistics, and improve customers’ shopping experience. The value of Amazon’s big data to
the company is multifaceted, and they all have a positive impact in different ways.

• Improve user experience

• Inventory optimization
• Improve the return on advertising
• Help to improve the service level
• Security

III. SIX PHASES OF BIG DATA

1. Data Generation:
This phase involves generating large amounts of data from user clicks, user searches,
shopping cart contents, product sales, return information, user reviews, user browsing
time, and more. Amazon’s Alexa Intelligent Assistant collects voice interaction data,
including voice commands, natural language conversations, and user requests.

2. Data Acquisition:
 Amazon collects data from its online marketplace, AWS cloud services, devices like
Alexa, and third-party sellers who use Amazon’s platform. When it comes to data
preprocessing, Amazon identifies and processes errors, missing values, and outliers in
the data. This may involve deleting invalid records, filling in missing data, correcting
inconsistent data, or correcting outliers to reasonable values. Data is often stored in
different formats, units, or structures. Amazon may need to convert the data into a
consistent format for analysis. Amazon also deletes redundant records. If the data
comes from different sources, Amazon needs to consolidate them into a consistent
data set to support comprehensive analysis. Amazon also evaluates the quality of the
data, including the accuracy, completeness, and consistency of the data, to ensure that
the data meets requirements.

3. Data Storage
Amazon employs a range of data storage technologies to manage the vast amounts of
data it has accumulated. This includes data lakes, databases, and distributed storage
systems to efficiently store structured and unstructured data. Amazon stores data in a
variety of ways, choosing different data storage solutions for different needs and uses.
Here are the main ways Amazon stores data:
Amazon S3 (Simple Storage Service): Highly scalable object storage service used to
store large- scale data.
Amazon RDS (Relational Database Service): Simplifies the setup, operation, and
scaling of relational databases in the cloud.
Amazon DynamoDB: Amazon DynamoDB is a managed NoSQL database service
for storing semi-structured and unstructured data.
Amazon Redshift: Amazon Redshift is a cloud data warehouse service designed
specifically for analytical work- loads.
AWS Data Lakes: Integrate multiple data types (structured, semi-structured, and
unstructured) into a single data store, facilitating centralized storage of data.

4. Data Analysis:
Amazon uses big data technologies such as Apache Hadoop and Apache Spark to
process and analyze large-scale data sets, which support batch and stream processing
analytics involving the use of distributed computing techniques to extract insights
from the collected data.
• Cluster analysis: Cluster analysis is used to group data into similar clusters. Amazon
uses this method to discover product categories, user behavior clustering, and more.
• Descriptive statistical analysis: Amazon uses descriptive statistical methods to
understand the basic characteristics of data, including mean, median, standard
deviation, percentile, etc., to help obtain information about data distribution and
central trends.
• Regression analysis: Amazon uses regression analysis to model relationships
between variables to understand how one or more independent variables affect
dependent variables. This can be used to predict sales trends, price elasticity, etc.
 • Time series analysis: For time series data, Amazon uses time series analysis methods
such as seasonal analysis and trend analysis to identify time patterns and trends.

5. Data Visualization or Interpretation

Amazon uses a variety of tools and techniques for data visualization to effectively
communicate data insights and insights to decision-makers and teams. Here are some of the
data visualization methods and tools Amazon may employ:

• Amazon QuickSight
• Tableau
• Jupyter Notebook

6. Decision Making

Amazon makes full use of big data to support decision-making, which is one of the keys to its
success. Here are the main ways Amazon uses big data for decision-making.

• User behavior analysis: Collect and analyze a large amount of user behavior data,
including purchase history, search queries, browsing history, shopping cart contents,
and more. This data helps understand customer needs and shopping habits to optimize
product recommendations, pricing strategies, and inventory management.
• Personalized recommendations: Using machine learning algorithms to provide
personalized product recommendations for each user. This enhances the shopping
experience and increases sales.
• Supply chain optimization: Including inventory management, order delivery, and
transportation route planning. This helps to provide faster delivery, lower costs, and
reduce inventory waste.
• Advertising and marketing: Analyzing the performance of advertising and
marketing campaigns to determine which channels and strategies are most effective.
• Product development and improvement: To guide the development of new products
and improve existing products. This helps meet customer needs and innovate.
• Predictive analytics: For sales trends and demand forecasting, improving advanced
machine learning algorithms to better predict what customers across the country will
need in order to have the right inventory in the right area at the right time.

IV. THE NECESSITY OF BIG DATA SOFTWARE

In Amazon’s operations, big data software plays an indispensable role in various critical
areas. These crucial domains encompass supply chain optimization, customer service and
feedback, market competition, and pricing strategies, inventory demand forecasting, and data-
driven advertising and promotion. All of these aspects rely on different big data tools and
technologies to ensure that Amazon can maintain its competitiveness in a fiercely competitive
market while providing an exceptional customer experience.
Supply chain optimization: Big data software such as Apache Hadoop, and Spark to
monitor inventory and order statuses in real-time, ensuring the timely delivery of products.

Customer service and feedback: Big data analytics tools like Elasticsearch and Logstash
are used to analyse customer feedback and service requests, helping Amazon gain a better
understanding of customer satisfaction, issue resolution speed, and product quality. Such
analysis assists Amazon in promptly improving its services, thus enhancing customer
satisfaction and increasing customer loyalty. Market competition and pricing strategy have
always been Amazon’s core concerns. Amazon employs big data software such as web
crawlers and data mining tools to collect and analyse competitors’ price and product
information. This helps Amazon adjust its own pricing strategies and product offerings to
maintain its competitiveness. Accurate inventory demand forecasting is vital for Amazon to
avoid overstock or understock situations. Time series analysis tools and machine learning
models assist Amazon in analysing sales trends, predicting future demands, optimizing
inventory management, and increasing efficiency. Furthermore, data-driven advertising and
promotion are essential tools for driving sales. Amazon’s advertising services and
promotional platforms rely on big data analysis to precisely target advertising audiences,
increasing ad conversion rates and return on investment (ROI).

1. Data Processing Aspect

Apache Hadoop: Big companies like Amazon, IBM, Google, and Microsoft have adopted
Hadoop. For instance, after Microsoft’s adoption, the overall cost of analysis and business
intelligence products decreased by an average of 21.9%. The Hadoop big data analytics
market is expected to grow at a compound annual growth rate of 13.0% during the forecast
period, increasing from $127.97 billion in 2020 to $235.26 billion in 2025.
Apache Spark: Due to Amazon’s intake of a substantial volume of structured and
unstructured data, Spark has advantages in handling iterative and interactive algorithms. Its
in-memory computing feature accelerates the processing of big data, and it supports various
programming languages, including Java, Scala, and Python.

Cloud Computing Platforms: AWS is the most trusted cloud computing provider, offering
not only outstanding cloud security but also exceptional cloud services. It provides elasticity
and scalability for data processing. This means that Amazon can dynamically adjust
computing resources according to the continuously growing data processing needs.

2. Big Data Analysis Aspect

Amazon QuickSight: It is tightly integrated into Amazon’s cloud computing platform,
enabling users to easily connect to data sources and create dashboards for real-time
monitoring of critical business metrics.

Tableau: Empowers Amazon’s data analysis teams to create interactive dashboards and
reports for intuitively presenting information from large datasets. This helps them promptly
identify patterns and trends within the data, supporting decision-making and strategic
planning.
QlikView: Amazon’s teams use QlikView to rapidly analyze large datasets to gain in- depth
insights, assisting them in making informed business decisions.

3. In Machine Learning Model Building

To derive predictive data and offer personalized product recommendations to each user based
on their purchase history, search records, and behavior, Amazon extensively employs
machine learning and artificial intelligence software. This not only enhances user experience
but also increases sales.

Amazon SageMaker: This comprehensive machine-learning platform streamlines the

process of model development and deployment, enabling Amazon to apply machine-learning
technologies more rapidly across various business areas.

Natural Language Processing (NLP): To analyze user reviews and feedback, understanding
the strengths and weaknesses of products. This helps improve product quality and provide
goods that better meet customer needs.

Amazon benefits from machine learning services to enhance its product search engine,
making it easier for users to find what they need. By analyzing metrics such as search
keywords, click-through rates, and conversion rates, Amazon continually optimizes the
ranking and relevance of search results. They also employ machine learning models to detect
potential fraudulent activities, such as credit card fraud and fake accounts. This helps protect
user and transaction security while reducing losses caused by fraud.

4. Data Storage and Data Security

Due to the vast amount of transaction data, user behavior data, and product information
generated by Amazon every day, effective data storage and data security are of paramount
importance, which is precisely what big data software helps address.
S3: This cloud storage solution offers scalability and elasticity, enabling Amazon to handle
the ever-growing volume of data without concerns about hardware infrastructure limitations.
For Amazon, the reliability of data storage and the security of data are of paramount
importance. Amazon’s cloud storage solution provides data redundancy and backup features
to ensure data persistence and availability. This means that even in the event of hardware
failures or other issues, data remains accessible and recoverable.

V. Key Obstacles for Big Data Solution

In the era of information explosion, the speed of information generation is increasing. A large
amount of information is generated in the world. In the past few years, big data has become
one of the most commonly used terms in fields such as software, industry, and finance. It is
more concerned with most fields and has begun to use big data for analysis and discovering
new value.
Amazon’s users come from all over the world, so its database is large and complex. With
millions of products, they recommend them to customers who have a strong intention to
purchase. Their algorithms have been well-experienced on YouTube and Netflix as well.
However, how to improve the purchase rate of users is a question that Amazon has been
considering and is very important for merchants on better service platforms. The
recommendation system algorithm they have researched successfully solves this problem.
Amazon’s recommendation algorithm was named by the editorial committee of the “IEEE
Internet Computing” magazine as a paper that can withstand the test of time during its 20th
anniversary celebration in 2017.
Amazon Web Services Lab conducts large-scale A/B testing (structured testing of variant
products) to achieve data-driven business decisions. “The world focuses on user-based
collaborative filtering. Users visit websites: Which users are similar to them? We turned it
around and found a different approach that has better scalability and quality characteristics
for online recommendations”.
This algorithm allows Amazon’s user base to discover the products they want more quickly
on the Amazon platform, and it can also make the products provided by merchants sell better.

Artificial intelligence has significant traction among top companies, with AI ranked as the no.
1 strategic technology, according to the 2018 Research Trends Survey. By 1 year, the ability
to use AI to enhance decision- making, reshape business models and ecosystems, and reshape
the customer experience will drive returns on digital initiatives. 2025% of organizations are
still gathering information to develop their AI strategy, while the rest are already making
progress in piloting or adopting AI solutions, according to a Gartner survey. Constantly
training machine learning on datasets is a challenge for Amazon’s SageMaker product, which
is a machine learning platform on Amazon’s cloud that performs machine learning on top of
large, evolving datasets, and Amazon hasn’t found a ready-to-use solution available in
academia or the tech community for the need to encounter five challenges:

• How to train incrementally and keep the freshness of the model:

– Support incremental model training.
– Ensure regular and cost-effective model updates.
– Address the challenge of constantly generated data.
– Balance training cost and model accuracy.
• Predictable Training Costs:
– Enable users to estimate training costs and run-time.
– Overcome challenges in predicting costs for scalable systems.
– Address irregular performance drops in high-dimensional models.
• Elasticity and Job Management:
– Provide elasticity for handling uneven workloads.
– Support pausing and resuming training jobs.
– Facilitate tasks like hyperparameter tuning.
– Handle situations with irregular compute bandwidth limitations.
• Handling Ephemeral Data:
 – Manage data streams that are not meant to be stored.
– Overcome challenges in data ingestion and learning from such data.
• Automation of Hyperparameter Optimization and Model Tuning:
– Automate hyperparameter optimization.
– Streamline model tuning.
– Reduce the complexity of these tasks, particularly in large-scale scenarios.
– Enhance cost-effectiveness and usability, even for non-expert users.

VI. CONCLUSION

Thanks to its early adoption of big data principles and their integration into its business
strategy, Amazon has expanded beyond its humble beginnings as an online library. It has also
made significant investments in big data analysis and the creation of exploration applications.

Additionally, the business has created a technological roadmap that addresses consumer
concerns, increases tracking capabilities, and safeguards payment, delivery, and shopping
processes, among other processes. All of these factors suggest that Amazon’s success in
spearheading e-commerce is correlated with its level of teamwork, degree of control over a
vast amount of data, and the kind of technological tools it uses to support its strategic vision.

According to the study’s findings, Amazon’s business strategy is focused on achieving low-
profit margins through price reductions, option diversification, and the creation of systems
that enhance the user experience. Furthermore, Amazon’s technical tools are distinguished by
their versatility, adaptability, and application integration; this has led to superior control over
data management and more thorough and efficient analysis of their customer data.

Through the creation of more polarized responsive outputs for customer concerns, the
incorporation of big data into Amazon’s business strategy has helped to generate high-added
value. With this current move, Amazon hopes to provide its combined experience to startups
by offering aggregation services, processing, exploration, and big data analytics.

VII. REFERENCES

[1] Mustapha Bouakel and Amina Zerbout. “Perspectives of big data analytics’ integration in
the business strategy of amazon, inc.,” in Big Data Analytics, pp. 201-220. Apple
Academic Press, 2021.
[2] Sajee Mathew and J Varia. “Overview of Amazon web services,” Amazon Whitepapers,
vol. 105, pp. 1-22, 2014.
[3] Andy Jassy, “2022-amazon-report,” 2023.
[4] Raghavendra Kumar Indla. “An overview on Amazon recognition technology,” 2021.
[5] Ignacio Bermudez, Stefano Traverso, Marco Mellia, and Maurizio Munafo. “Large scale
observation and analysis of amazon aws traffic,” Politecnico di Torino, Tech. Rep, 2012.
[6] Changhao Liu, Yu Zhang, Yuanbo Xu. “The Impact of Big Data on E-commerce: A Case
Study of Amazon,” 2024