Running Head: DATA ANALYSIS 1

Data analysis

University of the Cumberlands

Data science

04/12/20

Q- W Mapreduce, Endgame

Q1 Storage Costs
a. Explain the different types of storage on AWS
b. Compare the storage costs between AWS, Azure and Google Cloud Platform
Helpful links:
• http://www.enterprisestorageforum.com/storage-management/cloud-storage-pricing.html
• https://www.slideshare.net/rightscale/cloud-storage-comparison-aws-vs-azure-vs-google-vs-ibm
Q2 Big Data Architecture
Explain the Big Data Architecture in
a. Microsoft Azure
b. Amazon AWS
c. Google Cloud Platform GCP
Links
http://azure.microsoft.com/en-us/documentation/services/hdinsight/
https://aws.amazon.com/big-data/getting-started/
https://cloud.google.com/products/big-data/
Q3 How Uber uses Big Data
Review the Big Data/Machine Learning Tools used by Uber
https://www.forbes.com/sites/janakirammsv/2019/06/26/managing-machine-learning-models-the-uber-way/
#7e1cf82f4ae4
Q4 Netflix
How does Netflix use Machine Learning
https://becominghuman.ai/how-netflix-uses-ai-and-machine-learning-a087614630fe
 DATA ANALYSIS 2

Question 1

a) The different types of AWS storage

i. AWS S3 Standard is the multi-purpose collection of commonly used data. It is a simple

database class that offers high capacity, accessibility of object space for regularly used data.

ii. AWS S3 Standard is the type for long term applications, but it is considered to be less

accessed data.

iii. AWS S3_Glacier implies that once the S3 Lifecycle is set, the firm's data will be moved

immediately to another processing class with no adjustments. It's the most stable, the most

durable.

iv. AWS S3RR this is an Amazon S3 database alternative that allows users to preserve anti-

critical data at a low level of durability than the standard Amazon S3 storage. It offers a free

open content distribution platform that is permanently placed somewhere.

b) Comparison of the storage costs between Azure, AWS and Google cloud platform

The three will also be capable of reaching almost similar average estimated costs by AWS and

Azure based on their total sum in various ways (Tiwary, 2019). Azure pays a premium for a

virtual computer, while AWS sets a limit on processing. Google Cloud also imposes a service fee

but reduces the expense of Active Directory service, which only poses more concerns that need

to be addressed before the last decision is made.

Question 2.

a. Microsoft Azure

• The main objective is the development of big data is to track, evaluate, and handle specific

datasets that are too massive for the older dataset framework.
 DATA ANALYSIS 3

• Data connected to graphics rendering activities can be stored in a cloud server in various ways,

and massive databases could also be saved in large volumes. This store is called a data center.

Azure Files, Blob, and Lake Store, are storage solutions.

• Huge databases of big data are analyzed using graphics rendering for filtering, replication, and

quite often data collection. The retrieval of source files and the speaking and listening effects are

also used in the graphics rendering.

b. Amazon AWS

• The supplier gives the company the ability to park their information online. The data is stored on

computers that are wireless locations for businesses, and the information is stored and viewed

online. Cloud storage has been used where the company uses a huge amount of data space to

hold big data (Muangprathub, 2019). That's why they take computing services from providers,

including Amazon cloud services. Data stored in the data storage means that remote repositories

store data and have protection measures to protect information from unauthorized persons.

• Data owners can store information online and can encrypt files online from web servers. Such

techniques are valuable to businesses because their data are in a secure position, and no one can

steal the data. And the organization can also view and change data remotely at any period.

c) Google cloud platform for the GCP

• These days, big data linked to different sectors is rapidly increasing, with Wikipedia being

among the prominent software warriors on the industry, processing hundreds of Petabyte data.

• The development of sophisticated data collection and storage systems, as well as the ability to

collect and analyze data to make important hospitality management decisions, have also been

greatly enhanced.
 DATA ANALYSIS 4

Question 3.

Machine learning tools

1. Michelangelo

It is a machine learning framework that has integrated frameworks and resources from groups in

an edge-to-end method. It has made it easier for designers and data analysts around the business

to build and operate machine learning applications on a level.

2. Horovod

This is a scalable training program that abuses the GPUs. Uber access sourced Horovod and gave

it to the Software Framework for Machine intelligence, Big data and Deep Processing.

3.Ludwig

It is the most popular machine learning program in Uber. It's an open and free, machine learning

set of tools designed that allows the user to educate and check machine learning systems before

developing software.

Question 4.

Ways in which Netflix uses machine learning

1. Movie recommendations personalization, Netflix utilizes the experience of many other

subscribers with shared interests to suggest what you might be much more willing to watch next

to ensure you remain active and maintain your subscription service for much more.

2. Personalization and auto-generation, utilizing millions of video images from that of a current

film or display as a base point for the collection of photographs, Netflix transcribes these photos

and then rates each picture in an attempt to determine which thumbnails are most likely to result
 DATA ANALYSIS 5

in a button (Basu, 2019). Such estimates are based on what other people who are close to you

have focused on.

3. Location for movie production, utilizing information to help determine when and where to start

filming set the parameters of the planning and production movie demands. Remember, this is

more a software engineering automation issue than just an artificial intelligence system that

makes projections based on previous results.

References
Basu, S., Kaminski, J. W., Panepucci, E., Huang, C. Y., Warshamanage, R., Wang, M., & Wojdyla, J. A.

(2019). Automated data collection and real-time data analysis suite for serial synchrotron

crystallography. Journal of synchrotron radiation, 26(1), 244-252.

Muangprathub, J., Boonnam, N., Kajornkasirat, S., Lekbangpong, N., Wanichsombat, A., & Nillaor, P.

(2019). IoT and agriculture data analysis for smart farm. Computers and electronics in

agriculture, 156, 467-474.

Tiwary, S., Levy, R., Gutenbrunner, P., Soto, F. S., Palaniappan, K. K., Deming, L., ... & Cox, J. (2019).

High-quality MS/MS spectrum prediction for data-dependent and data-independent acquisition

data analysis. Nature methods, 16(6), 519.