Date post: | 27-Dec-2015 |
Category: |
Documents |
Upload: | oswin-bradley |
View: | 217 times |
Download: | 1 times |
Horizontal Segmentationand
MindGenomics
Veljko Milutinovic, Jakob Salom, Zoran Markovic, Zoran Ognjanovic, Nenad Korolija, Jasna Matic,
andHoward Moskowitz
1/65
Essence
• HorizontalSegmentation =A method to increase ScientificEffectiveness, etc.
• MindGenomics =A method to implement HorizontalSegmentation
• Applications =EducationalEffects + ScientificOutput, etc.
• Unique for MindGenomics, MG =A 2-stage approach synergizing general and individual
2/65
One Example is Worth 1000 Words
• The simpler – the better!• Example: Sales boosting in restaurant business• Stage#1(General): Developing (compile-time)• Stage#2(Individual): Typing (run-time)
Note! Can be used:• To maximize scientific/technologic development• To bring markets/peace (health/banking, …)
3/65
MindGenomics in a NutShell
Compile-time effort: e.g., Bringing people to a restaurant
4/65
MindGenomics in a NutShell
4. Generation of a Marketing Slogan3. Analysis2. Market Polling1. Building the MicroScience
Compile-time effort: e.g., Bringing people to a restaurant
4/65
MindGenomics in a NutShell
4. Generation of a Marketing Slogan3. Analysis2. Market Polling1. Building the MicroScience
Compile-time effort: e.g., Bringing people to a restaurant
4/65
MindGenomics in a NutShell
4. Generation of a Marketing Slogan3. Analysis2. Market Polling1. Building the MicroScience
Run-time effort: e.g., Treating people at arrival to the restaurant
Compile-time effort: e.g., Bringing people to a restaurant
4/65
MindGenomics in a NutShell
8. Building the Experience Base 7. Profit Analysis6. Targeted Action5. Customer Typing (at reservation time)
4. Generation of a Marketing Slogan3. Analysis2. Market Polling1. Building the MicroScience
Run-time effort: e.g., Treating people at arrival to the restaurant
Compile-time effort: e.g., Bringing people to a restaurant
4/65
MindGenomics in a NutShell
8. Building the Experience Base 7. Profit Analysis6. Targeted Action5. Customer Typing (at reservation time)
4. Generation of a Marketing Slogan3. Analysis2. Market Polling1. Building the MicroScience
Run-time effort: e.g., Treating people at arrival to the restaurant
Compile-time effort: e.g., Bringing people to a restaurant
4/65
MindGenomics in a NutShell
8. Building the Experience Base 7. Profit Analysis6. Targeted Action5. Customer Typing (at reservation time)
4. Generation of a Marketing Slogan3. Analysis2. Market Polling1. Building the MicroScience
Run-time effort: e.g., Treating people at arrival to the restaurant
Compile-time effort: e.g., e.g., Bringing people to a restaurant
4/65
1. Building the MicroScience
Silos (6)
Elements (6)
5/65
Combining one Silo and two Elements: 216. Optimal: 40 vignettes to answer! Max48The MoskowitzJacobs software creates the process: One vignette = 4 elements!Elements developed using psychodynamic principles: 5KEYS(Cognition, Emotion, Behavior, Sensates, Personality)
Example of a Vignette in a Set of 48 Vignettes : Q#1
6/65
Example of a Vignette in a Set of 48 Vignettes: Q#2=?
7/65
2. Market Polling
a. Internet: In USA about $5 (typ = 300)b. NeuroEconomy: Possible Future (MIT)c. Text/Media-Mining:
Future Research (ACM iASC-2013)d. Computer/Network-Acceleration:
Future Research (ACM ISCA-2013)
8/65
High-Tech Polling
9/65
The Brain Marked with Economically Relevant Areas
10/65
Advances in Media Mining,
IEEE/ACM, 2013
Aleksandar Mihajlovic, Vladisav Jelisavcic, Zoran Ognjanovic,Veljko Milutinovic, Zoran Markovic, and Hermann Maurer
11/65
Maxeler Mining from Social Networks
Kartelj 2012 12/65
3. Analysis: Linear Regression
13/65
3. Analysis: Linear Regression
13/65
3. Analysis: Linear Regression
13/65
4. Slogans
Slogan #1 Slogan #2
Slogan #3
14/65
4. Slogans
Slogan #1
Blackened LouisianaShrimps
Slogan #2
AFLALongLifeVaccine
Slogan #3
Special Music Wish
14/65
4. Slogans
Slogan #1
Blackened LouisianaShrimps
Slogan #2
AFLALongLifeVaccine
Slogan #3
Special Music Wish
The slogans will bring people to the restaurant!14/65
5. Customer TypingClassifying each reservation-making or just-arrived customer to a cluster!
?
The Major Tradeoff: Likelihood Maximization versus Cluster Count 15/65
5. Customer TypingClassifying each reservation-making or just-arrived customer to a cluster!
38% 20%
42%
15/65The Major Tradeoff: Likelihood Maximization versus Cluster Count
5. Customer TypingClassifying each reservation-making or just-arrived customer to a
cluster!
38%24%
20%12%
42% 64%
15/65The Major Tradeoff: Likelihood Maximization versus Cluster Count
5. Customer TypingClassifying each reservation-making or just-arrived customer to a
cluster!
38%24%6%
20%12%6%
42% 64%
88%
15/65The Major Tradeoff: Likelihood Maximization versus Cluster Count
6. Targeted Actions
What is your music wish?88%
16/65
7. Profit Analysis
1.
2.
e.g., LittleBay@London 3.
4.17/65
8. Machine Learning
Findingmissingpuzzleelements: For betterfutureestimate
18/65
Horizontal Segmentation
So far: NS = 1
How about: NS > 1
One cluster One Segment!
19/65
Horizontal Segmentation: Example
Before: Yogurt - 1000pcs@$1
20/65
Horizontal Segmentation: Example
Before: Yogurt - 1000pcs@$1
After:
Cherry - 1000pcs@$2
Bicycle - 1000pcs@$2 AFLA - 1000pcs@$2
? - 1000pcs@$2 Typing?20/65
Horizontal Segmentation: Example
Before: Yogurt - 1000pcs@$1
After:
Cherry - 1000pcs@$2
Bicycle - 1000pcs@$2 AFLA - 1000pcs@$2
? - 1000pcs@$2 Typing? Self-typing while the customer is at shelves! 20/65
Horizontal Segmentation: Example
Before: Yogurt - 1000pcs@$1
After:
Cherry - 1000pcs@$2
Bicycle - 1000pcs@$2 AFLA - 1000pcs@$2
Yni - 1000pcs@$2 Typing? MG-typing while the customer is at entry! 20/65
MindGenomics: Revisited
21/65
The Ice Point of View:What is MindGenomics? (1)
Why?
A method based on industrial psychology, mathematics, and DataMining,
which can be used to promote products, services, ideas, opinions ...
What is the essence?
If one wants to sell, one has to listen to the needs of the prospect!
22/65
What is MindGenomics? (2)
Company assessment!
1. A company with potentials is approached, and their capabilities are estimated, using methods from company auditing and financial engineering.
2. A product is selected with potentials for horizontal segmentation;the question is, how to figure out what the market needs!
23/65
What is MindGenomics? (3)Data gathering!
3. Based on the principles of industrial psychology, a set of N questions is generated, which penetrate deep into the needs of the target group of individuals.
4. A polling partner is engaged to approach the target audience, and to bring back the form filled in, which assumes the existing of a small motivation item.
24/65
41/6525/65
What is MindGenomics? (4)
Data analysis!
5. The results of the poll are analyzed, using a proprietary datamining software, based on sophisticated math, and recommendations are made
(for the essence and the form)
6. Crucial Step: Analysis!
26/65
43/6527/65
An Introduction to Linear Regression
Given a data set of n statistical units, a linear regression model assumes that the relationship between the dependent variable yi and the p-vector of regressors xi is linear. This relationship is modelled
through a disturbance term or error variable εi — an unobserved random variable
that adds noise to the linear relationship between the dependent variable and regressors. Thus the model takes the form:
where T denotes the transpose, so that xiTβ is the inner product between vectors
xi and β. Often these n equations are stacked together and written in vector form:
where:
28/65
Example
Figure 3: Example of simple linear regression, which has one independent variable!
29/65
What is MindGenomics? (5)
Process analysis!
7. The life time monitoring of the process is absolutely mandatory.
8. Applications span from science and engineering to humanities and arts!
What are the alternatives to polling?
9. SocialNetworksDataMining
10. NeuroEconomyMindMining
30/65
Definitions of Basic Terms
• IdeaMap• MindGenomics• AddressableMinds• RDE
31/65
IdeaMap = A Program
• Function: Running a study
• Input: – Demographic (facts for the polling agency)– Silos (concepts of interest for marketing)– Elements (concept related marketing slogans)
32/65
Intermediate Products 1. VIGNETTES 2. POLLING AGENCY INTERFACE
33/65After the vignettes are created, a human brain action might be needed before the poling starts!
.
1. VIGNETTES• Vignettes (screens with a scenario and questions): 40 or 48
Scenario (one silo and 3 to 4 elements from other silos)Questions (2 or 3): Think? Like? HowMuch?
• Scale: Each question on the scale ZERO to 9, or 1 to 9.• Important:
Questions are NEVER demographic! Questions are:
PSYCHOGRAPHIC (What do you like in this 50Y old picture?) BEHAVIOURISIC (Do you like the meeting to be casual?) MERCANTILISTIC (How much are you willing to pay or donate?) Patent: If 2 questions, then: B+M (more common) Patent: Combinations form consistency sets catch liars (acceptable justification for polling re-requests)
Patent: Others who praised/liked/bought the same book as you, also purchased ... (inserted into IdeaMap) Patent: Number of vignettes is study dependent (inserted into IdeaMap)
34/65
2. POLLING AGENCY INTERFACE
Several million people on file:Field House and Amazon Turk
Each person willing to read and click 96 times (typical 2 questions, in each one of 48 vignettes)
Requests and re-requests(demographic data must be precisely defined in re-requests)
35/65
Output
Elements ranked by utility values (each element gets a utility value assigned)
Human brain has to be activated at this point! Clusters of mind sets(to select three clusters manually)
36/65
MidGenomics = A ProcessThe scientific process of discovery:
a. How people thinkb. How people feelc. What are they ready to payd. [and in future, how they change]
For the following:a. Productb. Servicec. Idead. [HR]
Typing is a part of MG!
37/65
AddressableMinds = A Set of People
Set of people who belong toone of the categories discovered by MG!Only such people can be typed!
38/65
RDE = Rules Developed Experimentation
Alias: Development of MicroScienceGoal: Finding minds of given characteristicsExample: Peter Coleman (conflict resolution in Israel)
Scenaria: 1. $ to ALL
2. Finding what they like, and make it! 3. The best: Finding mindsets who like to cooperate online, and offering them a business opportunity :)
39/65
A RESTAURANT EXAMPLEStep#1: Talk and Google, to understand the essence!
Step#2: Make 6 Silos (e.g., music, food, wines, ambience, ...)
Step#3: Make 6 elements per Silo (slogans) This results in 36 marketing slogans, some good some bad: IdeaMap will rank them and will give each one a weight Those above a threshold: Good Those below the threshold: Bad
Examples: a. Food, wine, and opera - the best choice for the first date b. Select tunes to fit your wine bouquet c. ... These should be generated by a PsychoAnalyst! Long story!
40/65
MidGenomics = A ProcessStep#4: Submit all 36 to IdeaMap, after the MJ is ready for you. This means: An account was formed to which you log in. Upon creation of the account: Type 36 elements and press SUBMIT
Step#5: Get 48 vignettes, each one 2 questions; each question 1 to 9.
Step#6: Find the polling agency and give them demographic requests for 300
Step#7: Polling in which the polling agency (PA) does not see vignettes! PA sends only the URL to polling individuals (PI) PI = pollees or pollers PIs log in remotely and click 96 times.
41/65
42/65
43/65
44/65
Avg UK Fr GerSeg1Seg2 Base Size 287 73 126 88 42 31 Constant 21 8 32 16 12 3
A7Cereal bar with dried fruits and honey, no sugar added 8 7 8 9 -4 26
A8Soft and crispy cereal bar, dried fruits, seeds and nuts 11 9 13 9 4 22
A1 Crispy cereal bar 5 5 5 6 -1 14
A9 Premium bar, soft and crunchy with a lot of fruit and chocolate chunks 16 23 14 13 19 13
B3 Cereals from organic production 3 3 1 7 -2 12
A6Cereal bar with a soft fruit-yogurt filling 3 7 -7 13 -3 12
A3Crispy cereal bar with milk-chocolate coating 16 18 14 19 21 11
A4Chocolate bar with a filling of full grain biscuits and raisins 7 9 5 9 7 8
A2 Crispy cereal bar with dried fruits 1 7 -2 1 -3 7
Raw Data to Be Analyzed (e.g., Food Industry)
45/65
Mind Genomics Revealed Three Mind Sets (e.g., Segments for Teens Facing Hospital)
Seg1 Seg2 Seg3Base Size 43 55 59Constant 42 56 71
Seg 1 Medical staff always have a smile on their face 18 -8 -3Medical staff constantly checks up on teenage patients and insure their comfort 18 8 -4Medical staff genuinely tries to help patients which makes teenage patients feel special 16 -1 -7
Seg2 Medical staff always speaks the truth to the teenage patient no matter how traumatizing as it will help in the long run 5 15 -5Wittiness helps medical staff seem more human 2 15 -4Medical staff communicates and gives advice to teenage patients for their present and future lives 3 15 -4
Seg 3 Medical staff develop a teacher-student bond and help teenage patients who want to be medical staff themselves 3 -8 25Medical staff develops bond with teenage patient to make it easy for them to vent 7 0 14Medical staff develop a permanent friendship with teenage patient that carries on even after their release -3 -7 14 46/65
MidGenomics = A ProcessStep#8: Consistency check by IdeaMap If an inconsistency is caught, MJ re-requests N repeats, and specifies their demographics. If the PA assumes abuse (getting more than 300 w/o pay), MJ shows the inconsistencies!
Step#9: Getting reports, manual clustering, and selecting top 3 slogans for each cluster.
Step#10: Shipping top 3 slogans for each cluster to media!
Billboards, TV, Radio, Web, ... Maximal number of slogans
for a campaign: 5
47/65
TYPINGAxiom#1: The compile time axiom!
Type them before they get to the restaurant, e.g., when making the reservation, via web (interactive login) or phone (direct call or call center).
Axiom#2: After they come, they do not like to "push buttons", so if you have a "fresh-from-the-oven" offer, or the guest just dropped-in un-announced, the value exchange must be considerable (TABLET with 3 questions to select a free appetizer).
Final GOAL: Increased the spending and many happy returns :)
48/65
H2020: ICEs of EuropeMindMining = MediaMining + AlgorithmicPlethora
49/65
MoskowitzJacobs@HCNY
The ICE of USA
• Formed around NYU: Queens College• Goals: Teaching + Consulting + Research• Teaching: Army of promoters + Educating future experts• Consulting: e.g.,
Pfizer animal health care Private prenatal hospital rooms Student mind sets elaborated (self-esteem, control, concentration)
50/65
Vision for Europe: National ICE Growing MultiNational
The MindGenomics Process: Step#1: Getting incorporated
Step#2: The 50/65 memorandum/contract with the ICE of USA Step#3: The B2C infrastructure (.ppt explanation + customer contract)
The pitch of the explanation:
a. If MG is used as a one-step process: Better 5% to 25% or more
b. If MG is used as a two-step process: UNIQUE
c. Axioms of MG: Think? Like? How much ready to pay/donate?
51/65
Step#4: The First Project
From min 1Y to max 3Y
Price Structure = Expences + Fee(ICE)
Expenses: Polling + Travel (if any) + Specifics (if any)
Fee(ICE): MicroScience (one time) + IndividualTyping (one month) = $2K IndividualTyping (monthly) = $200 Sefte = 50% off
52/65
The MindGenomics Process
Step#5: LESSONS LEARNED ANALYZED!
Step#6: Careful selection of the next 9 projects!
53/65
The MindGenomics ProcessStep#7: Another 9 projects @ 50% discount
in cooperation with ICE(USA)
Step#8: Lecturing in front of few big prospects. Definition of a Big Prospect: Chain of Identical Units or Single Multi Variety Production
Examples: Dealerships (e.g., VW) or Monopolists (e.g., DELTA)
Step#9: Automating the process: Setting the production rules Setting the quality control rules Teaching the HRs Setting the quality control of quality controllers
54/65
Step#10: Lecturing in front of many small firms!
Local USA Chamber of Commerce Local National Chamber of Commerce Local Commercial Chambers of Managers
Pitch:
This is a game of large numbers! Worst seller with MG vs Best seller w/o MG = 23% better The gain is statistical (invisible becomes visible)
55/65
The MindGenomics ProcessStep#11: Learning How to Close the Sale in Big Noise!
Suggestions: Not to hire a marketing agency
a. One of co-owners (for big prospects)
b. N assistants under one of co-owners (for many small)
Step#12: Classification of Master Targets
Master targets are classified as:
a. Those having a problem e.g., NY Times selling subscription to college students
b. Those anticipating the problem e.g., Politicians before entering a campaign
56/65
The MindGenomics Process
Step#13: Education at a vector of universities!
a. Course structure (e.g., Freshmen's MATH 110 @ NYU): 1/3 Statistics (including linear regression) 1/3 The MG process 1/3 An MG class project
b. Invited lectures (e.g., Master and PhD Students)
Step#14: All child diseases cured!
57/65
Step#15: H2020 Proposal Preparation
Essence: Introduction to a vector of National ICE companies
Mechanism: 40 different partners from 40 different EU countries
Entrance Tickets: a. Have an educational course (on-site or via-skype) b. Have a paper published at a peer-review conference: Proof of understanding and building credentials c. Have a research idea compatible with the call d. Have a national ICE formed, as an affiliate of ICE.SRB: NGBT (no give before take) 50% stays locally 50% goes to ICE.SRB (for using the brand and technology) Half goes to ICE.USA Half stays at ICE.SRB Half stays with the majority owner Half goes to share-holders
Axiomatic issue: Unlimited IntraBorder Growth 58/65
Step#16: Proposal
Building on the top of Public Domain Improvements in 8 domain avenues!
Step#17: Patenting
Dr. Moskowitz holds (alone) 4 major MG patents
Step#18: Conferencing
e.g., VIPSI Brainstorming: What next?
Step#19: Journalization
Step#20: Exchange or JPMorgan 59/65
The MindGenomics Process
ESSENCE
Experimental design: Sets the vignettes
Linear regression: Deconstructs combinations,and detects lies as a byproduct
Clustering and discriminate function analysis: Create Qs for Typing
Empathy mining: Media based preprocessor - ongoing research!
60/65
OurPapers!Milutinovic, V., A Comparison of Suboptimal Detection Algorithms Applied to
the Additive Mix of Orthogonal Sinusoidal Signals, IEEE Transactions on Communicatiions, Vol. COM-36, No. 5, May 1988, pp. 538-543.
Milutinovic, V., Knezevic, P., Radunovic, B., Casselman, S., Schewel, J., Obelix Searches Internet Using Customer Data, IEEE COMPUTER, July 2000 (impact factor 2.205/2010).
Milutinovic, V., Cvetkovic, D., Mirkovic, J., Genetic Search Based on Multiple Mutation Approaches, IEEE COMPUTER, November 2000 (impact factor 2.205/2010).
Milutinovic, V., Ngom, A., Stojmenovic, I., STRIP --- A Strip Based Neural Network Growth Algorithm for Learning Multiple-Valued Functions, IEEE TRANSACTIONS ON NEURAL NETWORKS, March 2001, Vol.12, No.2, pp. 212-227 (impact factor 2.889/2010).
61/65
OurFP7s
BalCon ProSense ArtTreat HiPeac
62/65
NB:1. Using induction (datamining) create rules2. Using deduction (from global to specific) create questions from which one can deduce3. Example: Cooperation over children!
ENABLERS:1. Technology enables each individual to be addresses in a different way2. Condition: Modern behavior3. Example: Drop in who does not push buttons ;)
STRATEGIC LAYERS:1. SRB -> Europe2. JPM == World3. MAT == Universe
63/65
Raw Materials – Silos & Elements
Target Customers- Impact on Customer
Experience
INTERNET
IdeaMap™SURVEY
Identify topic area
ANALYZED SURVEY RESULTS → Mind Genomics
Total Sample
Seg 1 Self Driven Online
Banking Seekers
Seg 2 Technology/
High Security Seekers
Seg 3 Collaborative
Online Seekers
Seg 4 Personal Touch with Technology
Base Size: (267) (105) (50) (59) (53) Constant: 31 34 28 31 26
OC1 Connect online in 'real time' with a customer rep via instant messaging, voice over IP or video conferencing via your computer 0 2 2 5 -13
OC3 Faster loan application process…work in real time online with a loan officer 0 0 -8 12 -8
OC4 Our bank's customer service reps will help you browse and use our online banking services -1 0 -6 10 -10
OC2 Use our online tool to find and schedule at your convenience an online working session with an expert such as investment broker, insurance agent, and/or loan officer -3 1 -11 6 -12
ON3 We will answer all your requests in 'real time' by email, instant or text messaging 1 7 -11 3 -1
ON4 We offer 'On demand' status reports for services requests (e.g., loan application) delivered to you via e-mail, text or instant messaging
1 6 -4 -3 0ON2 No more paper mail... We will send you statements and images of
transactions securely by email -2 5 -17 -5 2ON1 We allow you to pay bills securely using your mobile devices (cell
phone, PDA, Blackberry, etc.) -4 2 -13 -9 -2
BR3 We offer a bank-issued smart card so we can recognize you entering the branch and process your needs faster 4 2 6 -3 13
BR4 Choose a secure eye or finger security scan to identify you immediately in-branch and at ATM 4 3 10 -1 3
BR2 We have the most secure biometric system that identifies you as you enter the branch so we can process your needs faster 2 -1 8 -5 6
BR1 We will recognize our customer's mobile phone signal when entering a branch so we can recommend appropriate bank products, promotions and special services -4 -3 -6 -11 2
Online Collaborative
Online Other
In-Branch Recognition
No more paper mail – we will send you copies of statements by secure e-mail
Securely manage your account by PDA, Internet of automated telephone
Our banks customer services reps will help browse & use our on-line services
Manage all your banking needs with a state of the art kiosk and be confident that live help is available if you need it
No more paper mail – we will send you copies of statements by secure e-mail
Securely manage your account by PDA, Internet of automated telephone
Our banks customer services reps will help browse & use our on-line services
Manage all your banking needs with a state of the art kiosk and be confident that live help is available if you need it
MARKET SEGMENTATION TYPING ENGINEMARKETING PHRASES
© 2011 iNovum LLC
Process: Develop the Microscience …
64/65
65/65The best ideas are coming from nature!