2.3 Methods for Big Data What is “Big Data”? Summarizing Big Data.
THEPOWEROF VISUALDATA... · The “Big Data” Trend 2 Source: IDC,2014. What makes big data big? 3...
41
21 st Century Marketing Conference THE POWER OF VISUAL DATA Shasha Lu
Transcript of THEPOWEROF VISUALDATA... · The “Big Data” Trend 2 Source: IDC,2014. What makes big data big? 3...
21st Century Marketing Conference
THE POWER OFVISUAL DATA
Shasha Lu
The “Big Data” Trend
2
Source: IDC,2014
What makes big data big?
3
3
Source: https://www.go-globe.com/blog/things-that-happen-every-60-seconds/ 2017
In 1 minute……
Youtube 700,000 hours of videos watched and 500 hours of videos uploaded
Facebook 243,000 photos uploaded and 70,000 hours of video content watched
Instagram 65,000 photos uploaded
Netflix 87,000 hours of videos watched
Tinder 1,000,000 swipes and 18,000 matches
Whatsapp 1,000,000 photos shared
… …
Which part is unstructured data? Why is it useful?
Strength of The ‘New’ Data
Which part is structured data? Why is it useful? What problems it may have?
Customer comments
Tweets
Forum discussions
Photo posts
Video posts
(Visual) content marketing
In-store shopping behavior
… …
5
Gain customer insights
Understand customer preference
Improve customer experience
Discover unmet needs
Generate new product/business ideas
Measure and optimize marketing effectiveness
Use face analytics to improve business practice
6
What can you tell from a face?
Demographics (age, gender, ethnicity, height, weight,…)?
Background (education, occupation, )?
Emotional states (joy, disgust, angry, …)?
Habits (smoking, drinking, stay up late, …)?
Traits (trustworthy, attractive, mature, …)?
Personality (extroversion, conservative, …)?
Behavioral tendency (risk seeking, leadership, competent… )?
Information from a Face
8
What can we do with these information?
Health and Lifespan
Lifespan Estimate: 84.6Lifespan Estimate:71.2
Can face reveal how long one will live?
How does it work?
A New Solution for Life Insurance
Source: https://demo.lapetussolutions.com/light/home
Online Dating
Finding Better Matches with Face
• Tom
• Male
• 32 years old
• 178cm
• Graduated from Cambridge
• …
• Mark
• Male
• 32 years old
• 178cm
• Graduated from Cambridge
• …
?
Challenges in Online Dating Screening Process
15
VS.
400 Survey Questions Facial Appearance-based
16
Face Preference-based Screening
Face Perception Framework
17
Local
Facial Feature
Global
Facial Feature
17
Ref: Yinghui Zhou, Shasha Lu, and Min Ding. “A Face Anonymity-Perceptibility Paradigm and An Application in Online Dating Industry”
The Matching Result
18
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
recommend top one recommend top four
without facialpreference
with facialpreference
Percentage
of correct
predictions
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
recommend top one recommend top four
without facialpreference
with facialpreference
Ref: Yinghui Zhou, Shasha Lu, and Min Ding. “A Face Anonymity-Perceptibility Paradigm and An Application in Online Dating Industry”
19
A New Solution for Online Dating
Do faces affect how a viewer reacts to an ad in the metrics that advertisers care
about?
20
Content Marketing
21
22
23
24
25
Is the effect large enough (and/or affect the right viewers) to warrant careful selection
of faces when constructing print ad?
26
27
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
Maxim
um
Ab
so
lute
Im
pro
vem
en
t
Product Categories
Maximum Absolute Percentage Improvement for High Rating Participants
Aad
Abrand
PI
Reference: Xiao, L. and M. Ding. (2014). Just the Faces: Exploring the Effects of Facial Features in Print Advertising. Marketing Science. 33(3):338-352.
Content Targeting
28Reference: Xiao, L. and M. Ding. (2014). Just the Faces: Exploring the Effects of Facial Features in Print Advertising. Marketing Science. 33(3):338-352.
Face Analytics in Content Marketing
29What can You do with face analytics?
30
Customer Insights from Video Analytics
31
Fashion Retailing
32
33
Reference: Lu, S., Xiao, L., & Ding, M. (2016). A video-based automated recommender (VAR) system for garments. Marketing Science, 35(3), 484-510.
Video-based Automated Recommender
Understanding Customer Preference from Video
34
Customer Feedback
35
“There are too many clothes displayed in a store. I don’t have the time or effort to carefully check every piece of clothing.
The system saves me a lot of time and
effort by finding what I like in several
rounds.”
“The system is like magic! It made recommendations on some clothes that I
would never pick myself, but when I
follow the recommendations and try
them on, they look amazingly good on
me.”
“I don’t trust a salesperson’s recommendations since their
recommendations are often very
subjective, and they make
recommendations based on what they
would like to sell, rather than how I look.”
▪ Field Test Result: intention to try on (95%),purchase (50%)
▪ Shopping Experience: similar(44%),very similar(20%)
▪ Easy-to-use: easy/very easy(56%), difficult（16%）
▪ Willing to use in retailing store: 89%
Visual Analytics in Fashion Industry
36
Where does visual data come from?
37
When to use?
38
Product Design
Content Marketing
Customer Analysis
The Future Battlefield for Marketers
39
Opportunity for actionable insights
The Eigenface Decomposition
41Reference: Xiao, L. and M. Ding. (2014). Just the Faces: Exploring the Effects of Facial Features in Print Advertising. Marketing Science. 33(3):338-352.
