Download - Big Data, Analytics and Data Science
Data
Big Data and Data ScienceDavid
Lambert
Agenda
• Introduction to Big Data
• The V’s
• What Businesses Need to Know
• Market Research
• Analytics
• Modeling
• Data Science
Introduction
Census 2013: Internet and Smartphone use
Introduction
140,000-190,000 job shortfall by 20184.4 million jobs in Big Data related by 2015 in US
Gartner 2012
Introduction
Introduction
Forbes 2014
What Is Data
Two Basic Types of Computer Data
• Structured
• Variable or Metric information that is easily accessible and shared between computers and databases. (time, date, location, user ID, file.pathway, sensor)
• Unstructured
• Information that is difficult to quantify such as text, emails, pictures, videos and other socially generated content and is beyond typical processing power.
What Is Data
Structured
Unstructured
What Is Data
Share of Global Internet Searches
1.2 Trillion searches per year 40,000 searches per second
How Much Data
http://www.internetlivestats.com
How Much Data
64kb for the 1969 moon launch
Why Now
Digital Storage Has Become Very Cheap Price for CPU Performance is Cheap
• Information Connectivity
• Databases • Machines • Employees • Customers • Products
1
Basic Corporate Interaction with Clients
Experiences are ‘Pushed’ on Consumers by companies. Companies advertise and market to what they want the consumer to think with little or no feedback.
2
Back and forth communication between ISOLATED users and producers.
Push-Pull marketing efforts evolve to gauge consumer experience. Exclusive Company/Customer Dialogue, costly Surveys, and limited Focus Groups were typical of this type of corporate relationship.
3
Consumer community connections and corporate relationships. User-to-user connections grew exponentially with the rise of social media platforms, like Facebook and the internet. Consumer experience shifted away from getting information from companies but rather gaining insight through advanced social webs. The general thought being that consumers-to-consumer interactions are less bias and convey more knowledge about an experience than an organizations. There Is an increased feeling of trust in dealing with a non-partisan opinion.
4 Cloud
Server
Cloud and Mobile system integration and infrastructure Through the development of advanced computer infrastructures, information growth was so rapid that it is referred to as an explosion of Big Data. The transition to smartphone use over standard computer use caused greater behavioral data to be captured by cloud services. Smartphones act as the most common gateway or remote to this advanced network computers.
Cloud
Service
5
Internet of things and User Connectivity The cost and benefit of connecting a product to a cloud or internet service captures so much value that companies everywhere are integrating systems to utilize this capability. Interactions between consumers and their products and product-to-product interactions will become increasingly more frequent and will boast a wave of new services to better integrate these systems into the consumer’s life.
Progression
Cloud
Cloud
Service Service Service
Service
PullPushWeb Cloud
Internet of Things
Server
Big Data
People are creating more Data
People, Products and Companies are creating ENORMOUS amounts of Data
Companies are creating more Data
Products are creating more Data
Big DataThe 3 Vs Chart
Innovation must be done more quickly and effectively due to this increase in
competition, availability of services and
dynamically changing technology.
Old business issues are still prevalent in
BD except the speed,
scope & tempo of
business offerings have
dramatically increased. Businesses
MUST take more control over their
service offerings.
The importance of
maintaining a
consistent and
authentic experience throughout all operations is
much more significant with the influence of BD.
Organizations must choose or
transition to
Attractive channels for
their objectives,
customers, and culture.
Greatest Challenge:Simply put, Big Data is an increase in the relationships between Hardware & Software, Products & Services, Customers & Employees within an organization or business.
The V’s
The Infinite V’s
1. Volume
2. Velocity
3. Variety
4. Variability
5. Veracity
6. Visualisation
7. Value
Data Capture
Data Cleaning
Data Preparation and Reporting
Data Marketing Pr
oces
s
Data Capture Data Cleaning Data Preparation Data Marketing
Process
What Businesses Should Know
What Businesses Should Know
https://infocus.emc.com/william_schmarzo/big-data-business-model-maturity-chart/
• Consultant at EMC Global Services
• Former Vice President of Advertising Analytics at Yahoo
Bill Schmarzo
Big Data Author and Blogger at EMC
*** Text is excerpted or paraphrased from several articles by Bill Schmarzo at his EMC blog.
53 41 2
What Businesses Should Know
1.Integrate (structured) Meta-Data with detailed
(unstructured) Behavioral Data to provide new metrics and new dimensions against which to monitor and
optimize key business processes.
Initial Big Data Focus: Optimize Internal Business Process
https://infocus.emc.com/william_schmarzo/the-4-ms-of-big-data/
There are three big data capabilities that organizations can leverage
to expand their business intelligence and data warehouse
investments to optimize versus just monitor.
2.Deploy predictive analytics to
uncover insights buried in the massive volumes of detailed structured and unstructured data. Having business users slice-and-dice the data to uncover insights does not work very well when dealing with terabytes and petabytes of data.
3. Leverage real-time (or low-latency)
data feeds to accelerate organizationalprocesses to identify and act upon business and market opportunities in a timely
manner.
What Businesses Should KnowUltimate Big Data Opportunity: Monetize External Customer Insights
As organizations advance along the maturity index, three organizational transformations take place to create
new monetization based upon the
customer, product and
market insights gleaned from the first three
phases of the maturity index
1. Organizations start to treat data as an asset, not a cost of business.
2. Organizations place formal processes to capture,
inventory, refine, and protect their analytics
as intellectual property (IP). Analytics, models, processes, etc.
3. Organizational confidence in making
decisions using data and analytics will grow.
Organizational investments in data, analytics, people, processes, and technology will be used to justify decision
making.
What Businesses Should Know
A SHIFT FROM SURVEYS TO INTERACTION
• Surveys are difficult to… • Scale • Incentivize customers • Assess validity • Ask appropriate questions for meaningful insight
• Interaction monitoring is more successful to… • Scale • Incentivize customers • Assess Validity • Monitor Appropriate behavior for meaningful insight
Interaction Collection = Data Mining
Market Research
Market Research
The platform or “skeleton” service initially starts out with very little information about that particular user and thus the experience for that user is typically unexciting or ambiguous.
User
Fundamental service strategy: solutions selling, system integration
Interaction
BehaviorKnowledge
Consumer Interaction Platform
As the User begins to interact with the platform, the user provides user-defined inputs to basic profile characteristics. User defined variables should be limited to varaibles that are not easily gathered from general usage information. Age, Gender, income or other basic characteristicsmay be identified.
Interaction
BehaviorKnowledge
Market Research
User
Fundamental service strategy:
solutions selling, system integration
Consumer Interaction Platform
General activity and interaction with the Service Platform gives enough behavioral data to give
an extrapolated rough sketch of a consumer’s behavioral profile. General usage data can include activity hours/per month, activity duration per use, typical activities performed during use.
Interaction
BehaviorKnowledge
Market Research
User
Fundamental service strategy:
solutions selling, system integration
Consumer Interaction Platform
Adoption of the service platform into the clients lifestyle gives a caricature-like view of a consumer, with
exaggerated likes and dislikes. The consumer engages with the
platform regularly and has customized it to their preferences and application.
Interaction
BehaviorKnowledge
Market Research
User
Fundamental service strategy:
solutions selling, system integration
Consumer Interaction Platform
Immersion into the platform by the user creates a sophisticated profile that can be leveraged to gain unprecedented insight and predictability into consumer behavior. The dedicated use of a service platform by a single user details an almost life-like portrait of the user. If the user is engaged in multiple channels the data becomes even more relevant as it is analyzed across different markets segments and applications.Interaction
BehaviorKnowledge
Market Research
User
Fundamental service strategy:
solutions selling, system integration
Consumer Interaction Platform
Low Customization
Profile Activation
Low Account Activity: A picture, couple friends
Active Member: Moderate Photos, Postings, Friendship
Prolonged Use: Lots of Engagement
Interaction
BehaviorKnowledge
Consolidated User Profile with
Behavior
Generate Value by analyzing behavior
Compile Similar Users into Segment Data Packages
Mobile Advertising and Proximity Analytics
Service Innovation and IT-‐Enablement through Behavioral Analytics. ExactTarget
Market ResearchConsumer Interaction Platform
Interaction
BehaviorKnowledge
Consolidated User Profile with
Behavior
Leverage Data (Generate Value)
Compile Similar Users into Segment Data Packages
Many Companies reflect only a rough sketch of their customers
Prolonged Use: Lots of Engagement
Market ResearchConsumer Interaction Platform
Interaction
Behavior
Consolidated User Profile
with Behavior
Leverage Data
Compile Similar Users into Segment Data
Packages
Many Companies reflect only a rough sketch of their customers
Prolonged Use: Lots of Engagement
Consumer Interaction Platform
Market Research
Marketing based on Behavior
It is important to note that this also applies to process.
Market Research
Analytics
Defining a system of metrics or variables to be monitored, compared and contrasted within that system to determine the PREDICTIVE power of specific variables to
specific outcomes.
Analytics
Key Performance Indicator (KPI)
• ROI
• Net Profit/Profit Margin
• Customer Acquisition Cost
Analytics
Advertising/Sales/Marketing
Interaction/Usage Sentiment
Brand Culture
Customer Acquisition
ROI
Conversion Rate
Social Media Engagement
Distinguishing Attributes or Characteristics
Behavioral Rituals and Norms
Attitudes
Lexicon
Positioning
User Experience
Where is the interaction
Usage Time
AnalyticsProduct/Corporate Relationship
Advertising/Sales/Marketing
Interaction/Usage Sentiment
Brand Culture
Customer Acquisition
ROM
Conversion Rate
Social Media Followers/Mentions
Distinguishing Attributes or Characteristics
Behavioral Rituals and Norms
Attitudes
Lexicon
Positioning
User Experience
Where is the interaction
Usage Time
• ROI• Net Profit/Profit Margin
• Customer Acquisition Cost• Lifetime Value of Customer
Analytics
Analytics
Analytics
Analytics
Analytics
Analytics
Analytics
Analytics
AnalyticsHow Google uses Analytics to drive Technology
AnalyticsHow Google uses Analytics to drive Technology
• Purchased by Google in 2006 • All Music videos except #8
• Music has a lot of replay value. Sharable
• Psy and Katy Perry Appear twice on the top 15
http://en.wikipedia.org/wiki/List_of_most_viewed_YouTube_videos
Google Adwords • Sunshine Dairy
Analytic Modeling
• Media Buy • Geography • Click Through-Rate • Impressions • Conversion Rate • Cost Per Click • Ad Position • Size • Media (Vid/Pic/Audio) • Lifetime Value of Customer
• Creative • Style (Color/Mood/Message) • Short/Long Term Branding
Metrics
Analytic ModelingAnalytic Modeling is using the identified variables to forecast possible outcomes for the future. These models will be compared to actual results to build better models that more accurately predict consumer influences and outcomes on the
market.
A & B TestingForecast Method
Analytic Modeling
A & B TestingForecast MethodExponential Smoothing
Simple Regression
Multiple Regression
Moving Averages
Substitution Forecasting
Hybrid Forecasting
Data ScienceA Data Scientist is a generic term for someone
who possess the ability to do a combination of jobs which include;
Data Development Large Data Statistics
Data Analyst
Data ScienceT - Shaped Skills
Data Science
What To LearnLearn to Code
Software Engineering
Algorithms & Data Structures
Visualization
Data Munging
Distributed Computing
Machine Learning
Supervised (SVM, Random Forest)
Unsupervised (K-means, LDA)
Validation, Model Comparison
What To Learn
Linear Algebra(Matrix Factorization)
Calculus (Integrals, Derivatives)
Distribution (Binomial, Poisson)
Summary Statistics (Mean, Variance, Std Dev.)
Multivariate Analysis
Mathematics Statistics
Learn Math and Stats
Where To Learn
Online Programs
datasciencemasters.org