Professional Documents
Culture Documents
Business Changing in
the Big Data Era
Content
Dynamic Characteristics of Big Data
Data-Driven
Proactive Pushing
Agile Sensing
Cross-Domain Fusion
Value Orientation
3 qiangwei@tsinghua 2016/07
Pre Information Age (Before 1960)
“Non-data” Environment
Little
Intuitions
Decision Support
Business data
Handwork
Decision problem: How many should we
produce in the next quarter?
Decision-making model: decisions are made mainly
according to managers' experiences and what his
sense on market.
Business Activities
4 qiangwei@tsinghua 2016/07
The Beginning of IT Era (Before 2000)
Small Data Environment
Transactional Processing
Much
Business Data
Decision problem: How many should we produce in the next quarter?
Decision process: managers ask analysts to sort out data over the last two
years and make a graph. Then they will make decisions based on the statistic
model and the possible market demand in the next quarter.
Business Activities
5 qiangwei@tsinghua 2016/07
Network Era (After 2000)
Massive Data Environment
Massive
Intelligent
Analysis
Transactional
Processing
Decision Support
IT-Enable Analysis
Decision problem: How much should we
Automation produce in the next quarter?
& Decision process: the automatic module
Networking embedded in the system will extract related
information, like inventory, production line,
staff on a regular basis, then automatically
conduct an optimization, and can prompt the
results proactively.
Business Activities Business Data
6 qiangwei@tsinghua 2016/07
Business Intelligence
Business Value
?
Massive Data
7 qiangwei@tsinghua 2016/07
A Case of NBA Sports
How can an NBA team win a game?
Traditional thinking: all-court press, cross and smear, quickly
pounce and the related techniques.
The new weapon of NBA coach: data mining tools
About 20 teams adopted IBM Advanced Scout System to
optimize their tactics combination.
The Magic used Scout and analyzed lineups, and finally won
in the competition with Miami.
8 qiangwei@tsinghua 2016/07
Data-Mined Victory of Orlando Magic
According to the analysis of the system, two guards Anfernee Hardaway
and Brian Shaw in the Magic starting lineup got -17 in the previous two
games, which means they lost 17 points more than what they got .
However, when Hardaway and backup guard Darrell Armstrong
combined, the Magic got +14 points.
In the next match, the Magic extended the playing time of Armstrong,
showing an effective output: Armstrong got 21 points, Hardaway got 42
points. The Magic won by 88:79. In the fourth match, Armstrong started
lineup first and beaten the Miami again.
9 qiangwei@tsinghua 2016/07
Dwyane Wade’s Scores of 2015
10 qiangwei@tsinghua 2016/07
Intelligent Analysis of Massive Sports Data
11 qiangwei@tsinghua 2016/07
Hypothesis-driven vs. Data-driven
Hypothesis-driven: Data-driven:
Experts/Users
Big Data Analysis
12 qiangwei@tsinghua 2016/07
Data-Driven
13 qiangwei@tsinghua 2016/07
Data Size In The Big Data Era
Data Size Meaning Illustrations
1 Bit 1 Bit
1 Byte 8 Bit An English character
1 Kilo Byte (KB) 1,024 Byte A small email
1 Mega Byte (MB) 1, 048,576 Byte A plain text of novel, a photo
1 Giga Byte (GB) 1,073,741,824 Byte 30-minute DVD film
1 Tera Byte (TB) 1,099,511,627,776 Byte 1~3 computer drives
1 Peta Byte (PB) 1,024 TB
1 Exa Byte (EB) 1,024 PB
1 Zeta Byte (ZB) 1,024 EB
1 YB 1,024 YB
1 DB 1,024 YB
1 NB 1,024 DB 1152921504606846976 ITB hard disks,
about 70 trillion kilograms
14 qiangwei@tsinghua 2016/07
Growing into Big Data…
15 qiangwei@tsinghua 2016/07
Web 1.0 Web 2.0 Web 3.0
Pull Push
16 qiangwei@tsinghua 2016/07
Characteristics of Big Data:4V
1. Volume 2. Variety
3. Veracity 4. Velocity
Different from previous eras, the Big Data Era records all
activities in the form of data.
17 qiangwei@tsinghua 2016/07
Generating Big Data
– A Case of Coronation of the Pope
Dead Data
Alive Data
18 qiangwei@tsinghua 2016/07
Proactive Pushing
19 qiangwei@tsinghua 2016/07
The Evolution of Business Decision Paradigm
20 qiangwei@tsinghua 2016/07
UGC (User Generated Content)
UGC
search(Google Trends, baidu index)
online comment(LeBook, taobao)
social BBS(microblog, blogger,
wechat)
professional network forum(baidu
stock forum)
online social network(QQ, wechat)
instagram(Tumblr)
video sharing and play(YouTube,
Youku)
transaction data(stock, futures)
expression of feeling(Emoji)
crowdsourcing(wikipedia, baidu
knows)
crowdfunding(PDAI, crowfunding)
……
21 qiangwei@tsinghua 2016/07
IoT (Internet of Things)
wearable healthy/medical devices
22 qiangwei@tsinghua 2016/07
UGC: Google Flu Trends - Predicting or Sensing?
Flu prediction with Google Trends
23 qiangwei@tsinghua 2016/07
Search “iPhone”
- Sales Forecasting = Intent Sensing?
Search
volume
24 qiangwei@tsinghua 2016/07
Uncertainty of UGC
One of the founders of quantum physics
Werner Heisenberg, who pointed out in a
paper in 1927, that particles measuring
position in the quantum world will
inevitably affect the velocity of the
particles. Measuring the position of the
particle will undoubtedly affect its speed.
This is called "uncertainty principle." Now,
the "uncertainty principle" can also be
applied on big data.
Over-prediction of Google Flu Trends
Public opinions are spreading over on Weibo
and Twitter
Subjectivity of UGC ?
25 qiangwei@tsinghua 2016/07
IoT: Healthcare Big Data Analytics
Guttag J. and Stultz C. collected the
ECG data of heart-disease patients.
They did some data mining and
found that the risks of three types
of abnormal ECG patients dying
from heart attack increased 1 to 2
times. It can identify the high-risk
patients who haven't been found
by the current branching screening
technology.
26 qiangwei@tsinghua 2016/07
IoT: Hu Huanyong Line
• In 1935, Hu Yong Hwan noted that Heihe
- Tengchong Line is the Chinese
population density boundary. 96% of the
population live in the southeast of line in
On the Distribution of the Chinese
Population.
http://im.qq.com/online/index.shtml
27 qiangwei@tsinghua 2016/07
IoT: Mobile Phone Users and Population Migration
28 qiangwei@tsinghua 2016/07
Agile Sensing
29 qiangwei@tsinghua 2016/07
Alibaba’s November 11 Festival
30 qiangwei@tsinghua 2016/07
A joke about Big Data
A pizzeria phone rang, the customer service staff picked up the phone.
Customer Service: XXX pizzeria. Hello, Mr. Chen, is there anything I can help you?
Customer: Hello, I want a ......, wow, how do you know who I am?
Customer Service: Mr. Chen, the CRM can prompt your profile on your incoming calls.
Customer: Oh, okay. I Want a seafood pizza .....
Customer Service: Mr. Chen, seafood pizza is not for you.
Customer: Why?
Customer Service: According to your medical records, your blood pressure and cholesterol
are high.
Customer: So ah. What's that you can recommend?
Customer Service: You can try our healthy low-fat pizza.
Customer: How do you know I would like to eat this kind of?
Customer Service: You borrowed a low fat healthy diet on last Monday in the Central
Library.
Customer: Good. Then I would like a family pizza king, how much?
31 qiangwei@tsinghua 2016/07
A joke about Big Data (ctd.)
Customer Service: 99 yuan, enough for your family of six persons. But your mother should
eat less, she made a heart bypass surgery last month, she is still in recovery.
Customer: ... Thanks for the reminder. Can I pay by credit card ?
Customer Service: Mr. Chen, I'm sorry. Please pay in cash, because your credit card has
been maxed out, still owes the bank 4807 yuan, and not including mortgage interest.
Customer: I will go to the nearby ATM withdrawals.
Customer Service: Mr. Chen, according to your records, you have exceeded withdrawal
limits today.
Customer: Well, sent pizza to my house directly. I have cash at home. How long will it take?
Customer: About 30 minutes. If you do not want to wait, you can ride here yourself, just for
15 minutes. And I can give you a 5 yuan discount if you do so!
Customer: Why?
Customer Service: Based on our GPS vehicle tracking system records. You have registered a
motorcycle with license plate number SB-748, you are currently riding the motorcycle near
NO.23rd of Jiefang East Road.
Customer:……
32 qiangwei@tsinghua 2016/07
Cross-Domain Fusion
33 qiangwei@tsinghua 2016/07
The Fifth V of Big Data - Value
Source: https://www.nissatech.com/the-key-ingredient-you-need-to-extract-big-value-out-of-big-data/
34 qiangwei@tsinghua 2016/07
Value Orientation
35 qiangwei@tsinghua 2016/07
The Big Data Business Modeling Paradigm
36 qiangwei@tsinghua 2016/07
Why Big Data now?
37 qiangwei@tsinghua 2016/07
Big Data can be collected!
38 qiangwei@tsinghua 2016/07
Big Data can be Processed!
Malaysia Interior Minister Zahid Hamidi told
parliament on Wednesday, the heavy burden
on International Criminal Police Organization
loosened immigration check. Zahid said that
40.2 million of lost passports database is so
big that it makes Malaysia database
management system paralyzed. The
International Criminal Police Organization
whose Headquartered is in Lyon, France, said
its database requires only 0.2 seconds to
show whether a passport has been listed as
stolen one.
39 qiangwei@tsinghua 2016/07
The Focuses of
Business Changing
40 qiangwei@tsinghua 2016/07
3 Principles to Create Business Value
Increasing Revenue
Acceleration
Cutting Cost
41 qiangwei@tsinghua 2016/07
Profit
=
Revenue – Cost
42 qiangwei@tsinghua 2016/07
Longtail Market
43 qiangwei@tsinghua 2016/07
Pareto Principle --80 / 20 Principle
Vilfredo Pareto
44 qiangwei@tsinghua 2016/07
Pareto Principle
80% of sales are created by 20 percent of the superior customers, a
direct corollary - the marketing focus is to explore, be kind to and keep
these excellent customers.
Why abandoning the other 80% customers. Because the sales made by
these customers is too small and can not meet the transaction costs for
marketing (bringing supply and demand together) .
For example, due to the limits of space and fund, a
supermarket can not put all its products of all brands on the
shelf. It can only sell their best-selling products. likewise, a
new vehicle design also pays more attention to mainstream
customers.
80% of the transaction cost is too high
to meet customer demand and can not
be offset by the low value benefits.
Therefore it can only be ignored -
ignoring the market!
45 qiangwei@tsinghua 2016/07
The Key of Matching Supply and Demand
- Transaction Cost
Matching
46 qiangwei@tsinghua 2016/07
Era of Zero Transaction Cost
The emergence and popularization of the network,
has made the transaction cost between the
Internet customers and the dealers approaching to
zero.
Its means there’s no need to give up any customer. You
can communicate with any customer effectively with very
low cost .
For example, 0 cost matching methods includes: fast
payment by Alipay, understanding user needs through
online reviews, analyzing the market situation through
search log, targeted marketing through mobile
advertising, locating consumer places via GPS, ...
47 qiangwei@tsinghua 2016/07
An Extreme Case of Long Tail
In Apple store there’s a new APP
published in May 2008, called the "I
Am Rich”, at $ 999. It was soon
removed off shelves for not following
Apple’s regulations. But within a short
period of being on shelves......
48 qiangwei@tsinghua 2016/07
Long tail of Amazon.com
Development History
The online bookstore was established in 1995 ( book is a standard
product )
In November of 1998, sales of book categories> 3 million kinds(Far
more than the traditional bookstores)
Through effective search, discounts, book reviews have gained
market recognition
Currently, Amazon's inventory turns up to 150 times (traditional
bookstores can just do 3 to 4 times)
In 2004, among books sold by Amazon, 57% of the varieties can’t be
found in physical bookstores like Barnes & Noble.
“In the long-tailed market of the Internet, 90% of the products can’t
be found in the traditional market. They contribute to 25% of sales and
25% of profits. At the same time, products that do not bring any profit
in the traditional market, occupying 8% in the total amount, account
for 25% of sales and 25% of profits in the long-tailed market. ”
49 qiangwei@tsinghua 2016/07
Today’s Amazon.com
• Amazon has thousands of patents of the logistics, with the
latest patent being “Anticipatory Shipping”.
50 qiangwei@tsinghua 2016/07
Characteristics of Building a Longtail Market
High speed search
Supply Intelligent Recommendation
Demand
Flexible / Customized
Social sharing
Customized needs
Manufacturing
Immersive Consumption
Virtual Shop
Optimized logistics
Marching Interactive customization
……
……
Precise Tracking
Real-time sensing
Mobile Payment
51 qiangwei@tsinghua 2016/07
Precisely "Match" the Longtail Market
The basis of precise matching - business intelligence
driven by big data.
Outbound Marketing Inbound Marketing
53 qiangwei@tsinghua 2016/07
Using UGC to Analyze Credit and Behaviors
In the afternoon of January 6th, 2015, the central bank
released the Notice about Preparation of Personal Credit
Investigation on its website. According to the notification,
companies including Ali Sesame Credit Management Ltd.,
Tencent Credit Investigation Co., Ltd., Ping An Shenzhen
Qianhai eight institutions credit Information Center Corp.,
Shenzhen PING AN Credit Investigation Co., Ltd., CCIS Co.,
Ltd. Intelligent Credit Co., Ltd., Kara Credit Management Co.,
Ltd. BEIJING SINOWAY CREDIT BUREAU need do
preparations about personal credit investigation services.
The preparation time is six months. This means that the
eight bodies may become China's first commercial credit
bureaus.
Credit data records Source: Alipay, TenPay, QQ chat records,
etc.
54 qiangwei@tsinghua 2016/07
Customers Deeply Involved in Customization
- Implicit Participation with IoT
EMC2 cooperated with German Audi,
designed optimized and customized
solutions based on the sensor data
(rather than self-reported data from
users) .
From the end of November in 2013, the iQIYI added a "Green Mirror"
video editing features. It is based on the operation of your video, such as
fast forwarding, rewinding , repeated playing, etc., to determine your
preferences automatically, then to generate Essentials. More than an
hour long“Dad,where we go " or "Happy Camp” ,that interspersed with a
lot of advertising, can be finished in watching on iQIYI for 30 minutes
without missing the highlights.
Green Mirror also showed the video content about Jimmy Lin taking care
of Cindy (Tian Liang’ s daughter), has become the highest rate of skipping
contents after opening and trailer in that program, among which the
fragment about Jimmy being indifferent when he saw Cindy was crying,
was skipped at the rate of 29.64%.
55 qiangwei@tsinghua 2016/07
Big Data Driven Total Lifecycle Marketing
Customer
design develop produce logistics store sales
service
56 qiangwei@tsinghua 2016/07
The Change of Marketing Ideas
From "selling product" to "selling customer“
The customized marketing department of Baidu, based on the database on 1 billions
customers’ online profile, sells the suitable target customers to advertisers.
Australian Airlines in 2014 had a loss of $ 2.8 billions AUD, but revenue of frequent flyers
department maintained growth for five consecutive years by sharing frequent flyers
information with major retailers.
57 qiangwei@tsinghua 2016/07
Crowdsourcing
Platform
58 qiangwei@tsinghua 2016/07
4 Enterprise System Development Strategies
high Business system low
Meal
development
buy mature commercial Eat at restaurant or
Purchasing system ; the suppliers provide
technological support have a takeout
c
Difficuilty
o the third party which is out of
the organization is responsible Employ a cook or a
Outsourcing
s for the construction and
maintanence of the system
part-time maid
t Support team inside the
Internal organization is responsible for
the development and Parents cook
Development maintenance of information
system
59 qiangwei@tsinghua 2016/07
Crowdsourcing
Crowdsourcing is a new form of production organization
brought by the Internet. It is a new business model that
companies use the Internet to assign work, find creative
points or solve technical problems. These organizations can
take advantage of the creativity and ability of voluntary
employees via controlling the Internet --- these voluntary
workers have the skills to complete the task, willing to use
their spare time to work and wanting a small remuneration
for their services, or no immediate reward, only to satisfy
the prospects getting more payment in the future,
especially for the software industry and the service sector,
this provides a new way of organizing labor.
60 qiangwei@tsinghua 2016/07
Searching Crowdsourcing - Human Flesh Search
(South China Tiger Incident)
On 3rd October , 2007, Shanxi farmer Zhou Zhenglong said that he took South China tiger
photos in Bashan; on the 12th of same month, the Provincial Forestry Department of Shanxi
held a press conference to show the South China tiger photos. After several hours, questions
about the real of "Tiger photos" appeared in the Seying Wuji BBS, then netizens constantly
questioned the picture from the light, camera angles, realistic paintings and other search
angles.
On 15th of November 2007, netizen named "Panzhihua xydz" said the tiger photos were very
similar to the pictures hanging in his house; the following days, netizens across the country
continued to report about finding " New Year tiger paintings", leading online discussions about
tiger photos. Voice thinking it’s fake gradually prevailed.
A netizen named "West undefeated" who played an important role in "abusing cat incident",
distinguished the trademark in lower left corner of "South China Tiger” carefully in Baidu South
China Ttiger BBS . He found a traditional Chinese character of "Dragon". West undefeated
then used the " New Year Dragon Painting", "Dragon mural", "Dragon wall painting" and other
keywords to search on the Internet.
As a result, he found the same trademark in the"Xinlong
wall painting" color printed and packaged by Lancaster
Company in Yiwu, Zhejiang...
You can‘t wrap fire in paper. On 29th of June, 2008, the
so-called "south China tiger photo" has been identified as
false picture finally, Zhou Zhenglong was arrested on
suspicion of fraud.
61 qiangwei@tsinghua 2016/07
Traffic Optimization Crowdsourcing ——
The “Real-Time Congestion Avoidance" between FM103.9 and
AMAP
FM103.9, Beijing Traffic Radio provided messaging platform
The drivers are notified through broadcast;
If you want to report a road traffic information, you can send a SMS to
the broadcast platform.
The radio host picks and reads relevant SMS.
The functions of AMAP "real-time congestion
avoidance"
On the one hand, through loading GPS and wireless
communications on the taxis in major cities ,
logistics cars and other vehicles on the industrial
operations, it can send and transmit the vehicle
travel time, speed, direction, coordinates and other
parameters information to a floating car dashboard
in center, and then comes to the traffic information
of road.
On the other hand, nearly ten million AMAP online
navigation users every day also gives a big data
about users’ traffic service.
62 qiangwei@tsinghua 2016/07
Crowdfunding – NoPhone project of KickStarter
September of 2015, the Kickstarter
website launched a NoPhone
project crowdfunding. NoPhone is
to response to the relief of phone
addicts, and their product is a
mobile phone-sized plastic plate.
63 qiangwei@tsinghua 2016/07
Information Crowdsourcing Platform
Farmeron provides farmers data tracking and analysis services like
Google Analytics. Farmers can use this software to record and track
their conditions of rearing livestock (fodder stocks, consumption and
spending, livestocks’ birth, death, milk and other information, as well
as information of farm income and expenditure ). Farmeron puts the
fragmented agricultural production documenting together. with
advanced analytical tools and reports, it can monitor the farm and its
production, helpful for farms to make independent and scientific
agricultural production plan.
Fragmented Supply
DOES Make Sense!
64 qiangwei@tsinghua 2016/07
Logistics Crowdsourcing
65 qiangwei@tsinghua 2016/07
AI Augmentation
66 qiangwei@tsinghua 2016/07
Deep Blue won Gary Kasparov (1997)
http://www.turbulence.org/spotlight/thinking/chess.html
http://www.bewitched.com/chess/
67 qiangwei@tsinghua 2016/07
Nowadays robotics
Google Car
68 qiangwei@tsinghua 2016/07
Non-power AI
Google Translation
Source: http://www.nytimes.com/2010/03/09/technology/09translate.html
69 qiangwei@tsinghua 2016/07
Non-power AI
IBM Watson
70 qiangwei@tsinghua 2016/07
Non-power AI
Computer can be artists!
Two paintings about Aaron Program developed by Harold
Cohen:
71 qiangwei@tsinghua 2016/07
Non-power AI
Computers can be an artist!
清平乐·黄菊 西江月·饮酒 点绛唇·佳人
72 qiangwei@tsinghua 2016/07
The Commercialization of AI
73 qiangwei@tsinghua 2016/07
Commercialization Features of AI
AI is able to learn quickly through the high-speed computation
(hardware acceleration + optimization algorithm), to achieve a
certain degree of intelligence.
For example, a production planning model needs 82 years with computer and
linear programming algorithm in 1988 ; but in 2003, the same model only
needs one minute, almost 43,000,000 times fast, i.e., 1000 times because of
the increase of CPU, 43,000 times because of the algorithmic optimization.
Features of AI
Global optimization (calm, no blind spots)
Feasible Solution Space, though very large!
Ultrahigh speed
Deep learning ability --- Innovation
speed up! speed up! speed up!
74 qiangwei@tsinghua 2016/07
AI Augmentation
• The human
But all these abilities are gradually
brain is able to
hosted by machines.
store data,
process
information,
extract
knowledge and
create wisdom.
75 qiangwei@tsinghua 2016/07
Challenge You or Augment You?
76 qiangwei@tsinghua 2016/07
Imagination
- Expand the Solution Space to Infinite!
You can
image
Break the red
light…
Drive off…
Flies to the high
latitude space…
……
77 qiangwei@tsinghua 2016/07
Motive Power -
Technology Innovation
78 qiangwei@tsinghua 2016/07
Profit
=
Revenue – Cost
Information Technology
79 qiangwei@tsinghua 2016/07
Horse Manure Crisis
in 1990s, the problems brought by the carriage:
Serious traffic congestion; unbearable noises of
wheels and hooves; many accidents (2 times of
today's accident accidents).
But the most intolerable thing is horse manure. A horse excretes manure
about 24 pounds per day. A total of about 200,000 horses’ manure weighs
about 5 million lbs a day in New York.
80 qiangwei@tsinghua 2016/07
The Resolution of Horse Manure Crisis
81 qiangwei@tsinghua 2016/07
Untimely Business Technology Innovation
Prof. Charles Babbage designed a theoretical model of
a mechanical computer in 1860.
83 qiangwei@tsinghua 2016/07