SlideShare a Scribd company logo
1 of 38
Download to read offline
Website Optimization
Problem and Its Solutions
Shuhei Iitsuka and Yutaka Matsuo
The University of Tokyo
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
A/B testing is powerful.
2
ref. How Obama Raised $60 million by Running a Simple Experiment

http://blog.optimizely.com/2010/11/29/how-obama-raised-60-million-by-running-a-simple-experiment/
8.3% 11.6%sign-up rate
$60M!
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Sample size is power.
3
Result
Result
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
See the wood first.
4
See the wood first. Search the neighbors.
Initialization Phase Local Search Phase
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Agenda
1. Related Studies
2. Website Optimization Problem
3. Proposed Testing Method
4. Experimental Results
5. Discussion & Conclusion
5
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Past Studies
Giants making profits by online testing with a large number of users.
6
1. Related Studies
However, how can we use it for smaller websites?
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Existing Testing Methods
7
A B
A B
A B
A B
B
A
A
B
A B
A B
A B
A B
A B
A/B Testing Full Factorial Design
Fractional Factorial Design Bandit Algorithm
1. Related Studies
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Agenda
1. Related Studies
2. Website Optimization Problem
3. Proposed Testing Method
4. Experimental Results
5. Discussion & Conclusion
8
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Expression of a Variation
A website variation can be denoted as a combination of elements.
9
=( , , )
Variation
→ The problem can be defined as a combinatorial optimization problem.
“GET
INVOLVED”
“CHANGE”
2. Website Optimization Problem
Website Variation:
Page Element:
x = (x1, · · · , xm)
xi 2 Vi
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Interaction with Users
10
p(y|x)f(x) ' E[y|x]
The evaluation value need to be estimated from the given feedback.
y ⇠ p(y|x)f(x) ' E[y|x] where
→ The evaluation function is estimated by the expected value.
2. Website Optimization Problem
xWebsite Variation
yUser Behavior
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Website Optimization Problem
Find the solution which satisfies the following equation.
11
x⇤
= arg max
x2X
E[y|x] s.t. y ⇠ p(y|x)
• maximizes the conditional expected value of the key metrics.
• is derived from the probability distribution.
2. Website Optimization Problem
x⇤
y
x⇤
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Local Search Solution
12
1. Initialization
2. Repeat until no improvement is made or all samples have been used.
2-1. Neighbor Solution Generation
2-2. Solution Move
X
x 2 X
X0
Neighbors(x)
x Move(x, X0
)
2. Website Optimization Problem
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Agenda
1. Related Studies
2. Website Optimization Problem
3. Proposed Testing Method
4. Experimental Results
5. Discussion & Conclusion
13
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Organization of Existing Testing Methods
14
Search Algorithm Technique
A/B Testing Local Search None
Full
Factorial Design
Brute-force Search None
Fractional
Factorial Design
Brute-force Search Linear Assumption
Bandit Algorithm Brute-force Search
Flexible Sample
Allocation
3. Proposed Testing Method
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Technique #1: Linear Assumption
15
Color Label Location
A B C L R
x = (x1, x2, x3)
3. Proposed Testing Method
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Technique #2: Flexible Sample Allocation
16
3. Proposed Testing Method
3.2% 2.4% 5.6% 1.6%
Expected Value
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Racing Algorithm
Another implementation of Flexible Sample Allocation.
17
3. Proposed Testing Method
ClickThroughRate
A B C D E
Variation
Remove
Adopt
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Overview of Proposed Method
18
Initialization Phase Local Search Phase
• Collects data from the entire
solution space.
• Estimates the optimal solution
with linear assumption.
• Start Local Search starting from
the estimated solution.
3. Proposed Testing Method
+ streamlined by flexible sample allocation
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Agenda
1. Related Studies
2. Website Optimization Problem
3. Proposed Testing Method
4. Experimental Results
5. Discussion & Conclusion
19
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Evaluation Experiments
1. Simulation Experiment / Artificial Problem
2. Simulation Experiment / Actual Large-scale Website
3. Practical Experiment / Actual Small-scale Website
20
4. Experimental Results
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Testing Methods
21
Method Initialization Local Search
BF
(Brute-force)
Random N/A
LA
(Linear Assumption)
Linear Assumption N/A
LS
(Local Search)
Random Local Search
LALS
(Linear Assumption +
Local Search)
Linear Assumption Local Search
LALS+
(LALS +

Racing Algorithm)
Linear Assumption +
Flexible Allocation
Local Search +
Flexible Allocation
Baseline
Proposal
4. Experimental Results
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Exp. #1: Simulation on Artificial Problems
22
Problem Evaluation Function Sample Size
#1 Linear Init. Only
#2 Linear Init. + Local Search
#3 Non-Linear Init. Only
#4 Non-Linear Init. + Local Search
f2(x) = x1 + x2 + x3 x4 x5 x6 x1x2 + N(0, 1)
f1(x) = x1 + x2 + x3 x4 x5 x6 + N(0, 1)
Problem Settings
Linear Evaluate Function
Non-Linear Evaluate Function
Nf(x)
xi 2 {0, 1, 2}
4. Experimental Results
Non-Linear Member
Noise
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Each method is evaluated by the accuracy of the estimated optimal solution.
23
Exp. #1 Results
Problem BF LA LS LALS LALS+
#1 (Linear/Small) 0.24 1.00 0.00 1.00 1.00
#2 (Linear/Large) 0.54 1.00 0.01 1.00 1.00
#3 (Non-Linear/Small) 0.26 0.14 0.01 0.22 0.22
#4 (Non-Linear/Large) 0.46 0.26 0.02 0.33 0.68
Baseline Proposal
Linear assumption works well with
the linear evaluation function.
Flexible sample allocation
boosts the local search.
4. Experimental Results
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Exp. #2: Simulation on a Large Website
• Actual large-scale website with 1000-10000 visiters/day.
• Key metrics: Ads Click-through Rate
• Evaluation function is simulated from the log (Mar 14-22, 2013)
24
A B C
Which one does maximize CTR?
SPYSEE http://spysee.jp
q(x) = 0.0640 + 0.0117xA 0.0067xB 0.0134xC
xi 2 {0, 1} (Apply the change or not)
4. Experimental Results
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Exp. #2 Results
25
0.25
0.50
0.75
1.00
0 10000 20000 30000
Sample Size n
Accuracy
Method
LALS+
LALS
LS
LA
BF
Average accuracy of each algorithm LA exhibits the best performance
because the evaluation function
is linear.
Our proposed methods succeeds
to start the local search from the
promising initial solution.
LALS+ can improve the
performance rapidly with the
flexible sample allocation.
Init. Local Search
4. Experimental Results
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Exp. #3: Practical Test on a Small Website
• Implemented our proposed method as an optimizer program.
• Actual small-scale website with hundreds of visitors/day.
• LS (Baseline) VS. LALS (Proposal)
• Key metric: Page views per session
26
Element Values
Thumbnail border width 0px, 5px
Thumbnail margin 0px, 5px, 10px
Thumbnail Size 100px, 200px, 300px
Thumbnail Shape square, circle
Imagerous* http://imagero.us
Tested Elements
4. Experimental Results
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Exp. #3: Results
• LALS reached a 57% higher
solution.

(t-test: 99% confidence)
• Our proposed method
functions as a practical
optimizer program with an
actual small-scale website.
27
Transition of the current solution and
the expected value.
ExpectedValueE[y|x]
0
2
4
6
8
Sample Size n
0 175 350 525 700
LS
LALS
4. Experimental Results
57%
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Agenda
1. Related Studies
2. Website Optimization Problem
3. Proposed Testing Method
4. Experimental Results
5. Discussion & Conclusion
28
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
From Bits to Atoms
29
Requirements
Each solution is expressed as
a combination of elements.
Reconfiguration cost is zero.

ex.) 3D printers
User feedback is observable.

ex.) Review website
5. Discussion & Conclusion
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Conclusion
• We formalized existing testing methods and a website
optimization problem.
• We proposed a new rapid testing method which works on small-
scale websites.
• We evaluated that our proposed method works on actual small-
scale websites.
30
5. Discussion & Conclusion
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Future Works
31
Make a Hypothesis
x 2 X
Define Metrics
f(x)
Explore the Solution
x⇤
= arg max
x2X
f(x)
We’ve tackled this!
Which key metrics we
need to focus for
effective experiments?
How do we define our
website as a set of
variables?
How can we automate
the generation of
candidates?
Website Optimization Process
5. Discussion & Conclusion
“Website Optimization Problem and Its Solutions (Paper ID:516)” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo
Shuhei Iitsuka, The University of Tokyo.
tushuhei.com
iitsuka@weblab.t.u-tokyo.ac.jp
Thank you for listening.
32
“Website Optimization Problem and Its Solutions (Paper ID:516)” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo
Appendix
33
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
X: Candidate Solutions
Y ← {} : Empty Set for Observed Data, n ← 0 : Number of Observations.
N_1: Sample Size for Initialization Phase, N_2: Sample Size for Local Search Phase.
FOR N_1 TIMES:
Y ← Observe(RandomChoice(X))
n++
x* ← LinearAssumption(Y)
WHILE n < N DO:
x’ ← GetNeighborSolution(x*, X)
FOR N_2 TIMES:
Y ← Observe(x’)
n++
x* ← Update(x*, x’, Y)
RETURN x*
34
Initialization
Local Search
3. Proposed Testing Method
+ Streamlined by
flexible allocation
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
DOE and Linear Assumption
• DOE (Design of Experiment) is used in traditional industries

which have huge cost to reconfigure the environment.
• Websites require no cost to change the parameters.

→ We can conduct random observation, then apply ANOVA to
estimate each element’s effect.
35
Design of Experiment:
Design beforehand.
Linear Assumption:
Random collection first.
Zero

Reconfiguration Cost
5. Discussion & Conclusion
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Webpage Segmentation
36
1ϕ
2ϕ
3ϕ
1
2
ϕ 2
2
ϕ
(a) (b)Cai, Deng, et al. Vips: a vision-based page segmentation algorithm. Microsoft technical report, MSR-TR-2003-79,
2003.
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Page Element Extraction
37
WELCOME!
JOIN NOW
Background: {WHITE, BLACK}
Button Color: {WHITE, BLACK}
Strong Interactive
Effect?
“Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015.
Bandit Algorithm
• ε-greedy
• ε: exploration, 1 - ε: exploitation
• Softmax: High expected value → High exploitation rate
• UCB1: Expected value + Freshness bonus
38
解 x をユーザに表示する確率 p(x) は式 2.1 によって表される。
p(x) =
exp(yx/τ)
x∈X exp(yx/τ)
(2.1)
呼ばれるパラメータであり、探究心の強さを表している。温度が非常に高い
わち τ → ∞ のときは解 x を選ぶ確率 p(x) は 1/|X| に収束するため、すべ
等の確率で選ばれることになる。逆に温度が低いときは yx が効き始めるた
価値の期待値が高い解が 1 に近い確率で選ばれるようになる。
CB1 ではこれまでに紹介したアルゴリズムとは異なり、ランダム性を用い
1 では基本的に評価値の期待値 yx が最も高い解を選ぶ戦略だが、解を選ん
じてボーナスが追加される。解 x ∈ X を表示した回数を tx とすると、解 x
ux は
ux = yx +
2 log( x∈X tx)
tx
、この UCB 値を最大にする解 x が選択される。
い解を優先的に表示することで実験による損失を免れている。簡単に実
、解の期待値に関わらず探求または活用を選択するため、期待値に大き
でも期待値が低い解を選んでしまう可能性がある。
max アルゴリズムでは、解の評価値の期待値に応じて表示する確率を
空間を X、観測データから算出される解 x ∈ X の評価値の期待値を yx
をユーザに表示する確率 p(x) は式 2.1 によって表される。
p(x) =
exp(yx/τ)
x∈X exp(yx/τ)
(2.1)
れるパラメータであり、探究心の強さを表している。温度が非常に高い
τ → ∞ のときは解 x を選ぶ確率 p(x) は 1/|X| に収束するため、すべ
確率で選ばれることになる。逆に温度が低いときは yx が効き始めるた
の期待値が高い解が 1 に近い確率で選ばれるようになる。
ではこれまでに紹介したアルゴリズムとは異なり、ランダム性を用い

More Related Content

More from Shuhei Iitsuka

Online and offline handwritten chinese character recognition a comprehensive...
Online and offline handwritten chinese character recognition  a comprehensive...Online and offline handwritten chinese character recognition  a comprehensive...
Online and offline handwritten chinese character recognition a comprehensive...Shuhei Iitsuka
 
Inferring win–lose product network from user behavior
Inferring win–lose product network from user behaviorInferring win–lose product network from user behavior
Inferring win–lose product network from user behaviorShuhei Iitsuka
 
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会Shuhei Iitsuka
 
Generating sentences from a continuous space
Generating sentences from a continuous spaceGenerating sentences from a continuous space
Generating sentences from a continuous spaceShuhei Iitsuka
 
ウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システムウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システムShuhei Iitsuka
 
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...Shuhei Iitsuka
 
Machine learning meets web development
Machine learning meets web developmentMachine learning meets web development
Machine learning meets web developmentShuhei Iitsuka
 
Python と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめるPython と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめるShuhei Iitsuka
 
リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門Shuhei Iitsuka
 
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...Shuhei Iitsuka
 
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web DataAsia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web DataShuhei Iitsuka
 
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ RuleShuhei Iitsuka
 
UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話Shuhei Iitsuka
 
ウェブサイトで収益を得る
ウェブサイトで収益を得るウェブサイトで収益を得る
ウェブサイトで収益を得るShuhei Iitsuka
 
HTML で自己紹介ページをつくる
HTML で自己紹介ページをつくるHTML で自己紹介ページをつくる
HTML で自己紹介ページをつくるShuhei Iitsuka
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おうShuhei Iitsuka
 
ウェブサービスの企画とデザイン
ウェブサービスの企画とデザインウェブサービスの企画とデザイン
ウェブサービスの企画とデザインShuhei Iitsuka
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おうShuhei Iitsuka
 
第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会 第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会 Shuhei Iitsuka
 
かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう Shuhei Iitsuka
 

More from Shuhei Iitsuka (20)

Online and offline handwritten chinese character recognition a comprehensive...
Online and offline handwritten chinese character recognition  a comprehensive...Online and offline handwritten chinese character recognition  a comprehensive...
Online and offline handwritten chinese character recognition a comprehensive...
 
Inferring win–lose product network from user behavior
Inferring win–lose product network from user behaviorInferring win–lose product network from user behavior
Inferring win–lose product network from user behavior
 
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
 
Generating sentences from a continuous space
Generating sentences from a continuous spaceGenerating sentences from a continuous space
Generating sentences from a continuous space
 
ウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システムウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システム
 
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
 
Machine learning meets web development
Machine learning meets web developmentMachine learning meets web development
Machine learning meets web development
 
Python と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめるPython と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめる
 
リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門
 
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
 
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web DataAsia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
 
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
 
UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話
 
ウェブサイトで収益を得る
ウェブサイトで収益を得るウェブサイトで収益を得る
ウェブサイトで収益を得る
 
HTML で自己紹介ページをつくる
HTML で自己紹介ページをつくるHTML で自己紹介ページをつくる
HTML で自己紹介ページをつくる
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おう
 
ウェブサービスの企画とデザイン
ウェブサービスの企画とデザインウェブサービスの企画とデザイン
ウェブサービスの企画とデザイン
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おう
 
第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会 第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会
 
かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう
 

Recently uploaded

Identifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population MeanIdentifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population MeanMYRABACSAFRA2
 
Predicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdfPredicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdfBoston Institute of Analytics
 
GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]📊 Markus Baersch
 
Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 2Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 217djon017
 
办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degree
办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degree办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degree
办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degreeyuu sss
 
Semantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxSemantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxMike Bennett
 
Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...
Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...
Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...Boston Institute of Analytics
 
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理e4aez8ss
 
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degreeyuu sss
 
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024thyngster
 
How we prevented account sharing with MFA
How we prevented account sharing with MFAHow we prevented account sharing with MFA
How we prevented account sharing with MFAAndrei Kaleshka
 
detection and classification of knee osteoarthritis.pptx
detection and classification of knee osteoarthritis.pptxdetection and classification of knee osteoarthritis.pptx
detection and classification of knee osteoarthritis.pptxAleenaJamil4
 
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...Amil Baba Dawood bangali
 
Conf42-LLM_Adding Generative AI to Real-Time Streaming Pipelines
Conf42-LLM_Adding Generative AI to Real-Time Streaming PipelinesConf42-LLM_Adding Generative AI to Real-Time Streaming Pipelines
Conf42-LLM_Adding Generative AI to Real-Time Streaming PipelinesTimothy Spann
 
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...ssuserf63bd7
 
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default  Presentation : Data Analysis Project PPTPredictive Analysis for Loan Default  Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPTBoston Institute of Analytics
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsVICTOR MAESTRE RAMIREZ
 
DBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfDBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfJohn Sterrett
 
Top 5 Best Data Analytics Courses In Queens
Top 5 Best Data Analytics Courses In QueensTop 5 Best Data Analytics Courses In Queens
Top 5 Best Data Analytics Courses In Queensdataanalyticsqueen03
 

Recently uploaded (20)

Identifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population MeanIdentifying Appropriate Test Statistics Involving Population Mean
Identifying Appropriate Test Statistics Involving Population Mean
 
Predicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdfPredicting Salary Using Data Science: A Comprehensive Analysis.pdf
Predicting Salary Using Data Science: A Comprehensive Analysis.pdf
 
GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]
 
Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 2Easter Eggs From Star Wars and in cars 1 and 2
Easter Eggs From Star Wars and in cars 1 and 2
 
办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degree
办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degree办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degree
办美国阿肯色大学小石城分校毕业证成绩单pdf电子版制作修改#真实留信入库#永久存档#真实可查#diploma#degree
 
Semantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxSemantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptx
 
Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...
Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...
Data Analysis Project : Targeting the Right Customers, Presentation on Bank M...
 
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
科罗拉多大学波尔得分校毕业证学位证成绩单-可办理
 
Deep Generative Learning for All - The Gen AI Hype (Spring 2024)
Deep Generative Learning for All - The Gen AI Hype (Spring 2024)Deep Generative Learning for All - The Gen AI Hype (Spring 2024)
Deep Generative Learning for All - The Gen AI Hype (Spring 2024)
 
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
毕业文凭制作#回国入职#diploma#degree澳洲中央昆士兰大学毕业证成绩单pdf电子版制作修改#毕业文凭制作#回国入职#diploma#degree
 
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
Consent & Privacy Signals on Google *Pixels* - MeasureCamp Amsterdam 2024
 
How we prevented account sharing with MFA
How we prevented account sharing with MFAHow we prevented account sharing with MFA
How we prevented account sharing with MFA
 
detection and classification of knee osteoarthritis.pptx
detection and classification of knee osteoarthritis.pptxdetection and classification of knee osteoarthritis.pptx
detection and classification of knee osteoarthritis.pptx
 
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
NO1 Certified Black Magic Specialist Expert Amil baba in Lahore Islamabad Raw...
 
Conf42-LLM_Adding Generative AI to Real-Time Streaming Pipelines
Conf42-LLM_Adding Generative AI to Real-Time Streaming PipelinesConf42-LLM_Adding Generative AI to Real-Time Streaming Pipelines
Conf42-LLM_Adding Generative AI to Real-Time Streaming Pipelines
 
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
Statistics, Data Analysis, and Decision Modeling, 5th edition by James R. Eva...
 
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default  Presentation : Data Analysis Project PPTPredictive Analysis for Loan Default  Presentation : Data Analysis Project PPT
Predictive Analysis for Loan Default Presentation : Data Analysis Project PPT
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business Professionals
 
DBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdfDBA Basics: Getting Started with Performance Tuning.pdf
DBA Basics: Getting Started with Performance Tuning.pdf
 
Top 5 Best Data Analytics Courses In Queens
Top 5 Best Data Analytics Courses In QueensTop 5 Best Data Analytics Courses In Queens
Top 5 Best Data Analytics Courses In Queens
 

Website Optimization Problem and Its Solutions

  • 1. Website Optimization Problem and Its Solutions Shuhei Iitsuka and Yutaka Matsuo The University of Tokyo
  • 2. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. A/B testing is powerful. 2 ref. How Obama Raised $60 million by Running a Simple Experiment
 http://blog.optimizely.com/2010/11/29/how-obama-raised-60-million-by-running-a-simple-experiment/ 8.3% 11.6%sign-up rate $60M!
  • 3. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Sample size is power. 3 Result Result
  • 4. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. See the wood first. 4 See the wood first. Search the neighbors. Initialization Phase Local Search Phase
  • 5. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Agenda 1. Related Studies 2. Website Optimization Problem 3. Proposed Testing Method 4. Experimental Results 5. Discussion & Conclusion 5
  • 6. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Past Studies Giants making profits by online testing with a large number of users. 6 1. Related Studies However, how can we use it for smaller websites?
  • 7. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Existing Testing Methods 7 A B A B A B A B B A A B A B A B A B A B A B A/B Testing Full Factorial Design Fractional Factorial Design Bandit Algorithm 1. Related Studies
  • 8. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Agenda 1. Related Studies 2. Website Optimization Problem 3. Proposed Testing Method 4. Experimental Results 5. Discussion & Conclusion 8
  • 9. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Expression of a Variation A website variation can be denoted as a combination of elements. 9 =( , , ) Variation → The problem can be defined as a combinatorial optimization problem. “GET INVOLVED” “CHANGE” 2. Website Optimization Problem Website Variation: Page Element: x = (x1, · · · , xm) xi 2 Vi
  • 10. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Interaction with Users 10 p(y|x)f(x) ' E[y|x] The evaluation value need to be estimated from the given feedback. y ⇠ p(y|x)f(x) ' E[y|x] where → The evaluation function is estimated by the expected value. 2. Website Optimization Problem xWebsite Variation yUser Behavior
  • 11. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Website Optimization Problem Find the solution which satisfies the following equation. 11 x⇤ = arg max x2X E[y|x] s.t. y ⇠ p(y|x) • maximizes the conditional expected value of the key metrics. • is derived from the probability distribution. 2. Website Optimization Problem x⇤ y x⇤
  • 12. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Local Search Solution 12 1. Initialization 2. Repeat until no improvement is made or all samples have been used. 2-1. Neighbor Solution Generation 2-2. Solution Move X x 2 X X0 Neighbors(x) x Move(x, X0 ) 2. Website Optimization Problem
  • 13. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Agenda 1. Related Studies 2. Website Optimization Problem 3. Proposed Testing Method 4. Experimental Results 5. Discussion & Conclusion 13
  • 14. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Organization of Existing Testing Methods 14 Search Algorithm Technique A/B Testing Local Search None Full Factorial Design Brute-force Search None Fractional Factorial Design Brute-force Search Linear Assumption Bandit Algorithm Brute-force Search Flexible Sample Allocation 3. Proposed Testing Method
  • 15. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Technique #1: Linear Assumption 15 Color Label Location A B C L R x = (x1, x2, x3) 3. Proposed Testing Method
  • 16. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Technique #2: Flexible Sample Allocation 16 3. Proposed Testing Method 3.2% 2.4% 5.6% 1.6% Expected Value
  • 17. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Racing Algorithm Another implementation of Flexible Sample Allocation. 17 3. Proposed Testing Method ClickThroughRate A B C D E Variation Remove Adopt
  • 18. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Overview of Proposed Method 18 Initialization Phase Local Search Phase • Collects data from the entire solution space. • Estimates the optimal solution with linear assumption. • Start Local Search starting from the estimated solution. 3. Proposed Testing Method + streamlined by flexible sample allocation
  • 19. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Agenda 1. Related Studies 2. Website Optimization Problem 3. Proposed Testing Method 4. Experimental Results 5. Discussion & Conclusion 19
  • 20. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Evaluation Experiments 1. Simulation Experiment / Artificial Problem 2. Simulation Experiment / Actual Large-scale Website 3. Practical Experiment / Actual Small-scale Website 20 4. Experimental Results
  • 21. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Testing Methods 21 Method Initialization Local Search BF (Brute-force) Random N/A LA (Linear Assumption) Linear Assumption N/A LS (Local Search) Random Local Search LALS (Linear Assumption + Local Search) Linear Assumption Local Search LALS+ (LALS +
 Racing Algorithm) Linear Assumption + Flexible Allocation Local Search + Flexible Allocation Baseline Proposal 4. Experimental Results
  • 22. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Exp. #1: Simulation on Artificial Problems 22 Problem Evaluation Function Sample Size #1 Linear Init. Only #2 Linear Init. + Local Search #3 Non-Linear Init. Only #4 Non-Linear Init. + Local Search f2(x) = x1 + x2 + x3 x4 x5 x6 x1x2 + N(0, 1) f1(x) = x1 + x2 + x3 x4 x5 x6 + N(0, 1) Problem Settings Linear Evaluate Function Non-Linear Evaluate Function Nf(x) xi 2 {0, 1, 2} 4. Experimental Results Non-Linear Member Noise
  • 23. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Each method is evaluated by the accuracy of the estimated optimal solution. 23 Exp. #1 Results Problem BF LA LS LALS LALS+ #1 (Linear/Small) 0.24 1.00 0.00 1.00 1.00 #2 (Linear/Large) 0.54 1.00 0.01 1.00 1.00 #3 (Non-Linear/Small) 0.26 0.14 0.01 0.22 0.22 #4 (Non-Linear/Large) 0.46 0.26 0.02 0.33 0.68 Baseline Proposal Linear assumption works well with the linear evaluation function. Flexible sample allocation boosts the local search. 4. Experimental Results
  • 24. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Exp. #2: Simulation on a Large Website • Actual large-scale website with 1000-10000 visiters/day. • Key metrics: Ads Click-through Rate • Evaluation function is simulated from the log (Mar 14-22, 2013) 24 A B C Which one does maximize CTR? SPYSEE http://spysee.jp q(x) = 0.0640 + 0.0117xA 0.0067xB 0.0134xC xi 2 {0, 1} (Apply the change or not) 4. Experimental Results
  • 25. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Exp. #2 Results 25 0.25 0.50 0.75 1.00 0 10000 20000 30000 Sample Size n Accuracy Method LALS+ LALS LS LA BF Average accuracy of each algorithm LA exhibits the best performance because the evaluation function is linear. Our proposed methods succeeds to start the local search from the promising initial solution. LALS+ can improve the performance rapidly with the flexible sample allocation. Init. Local Search 4. Experimental Results
  • 26. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Exp. #3: Practical Test on a Small Website • Implemented our proposed method as an optimizer program. • Actual small-scale website with hundreds of visitors/day. • LS (Baseline) VS. LALS (Proposal) • Key metric: Page views per session 26 Element Values Thumbnail border width 0px, 5px Thumbnail margin 0px, 5px, 10px Thumbnail Size 100px, 200px, 300px Thumbnail Shape square, circle Imagerous* http://imagero.us Tested Elements 4. Experimental Results
  • 27. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Exp. #3: Results • LALS reached a 57% higher solution.
 (t-test: 99% confidence) • Our proposed method functions as a practical optimizer program with an actual small-scale website. 27 Transition of the current solution and the expected value. ExpectedValueE[y|x] 0 2 4 6 8 Sample Size n 0 175 350 525 700 LS LALS 4. Experimental Results 57%
  • 28. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Agenda 1. Related Studies 2. Website Optimization Problem 3. Proposed Testing Method 4. Experimental Results 5. Discussion & Conclusion 28
  • 29. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. From Bits to Atoms 29 Requirements Each solution is expressed as a combination of elements. Reconfiguration cost is zero.
 ex.) 3D printers User feedback is observable.
 ex.) Review website 5. Discussion & Conclusion
  • 30. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Conclusion • We formalized existing testing methods and a website optimization problem. • We proposed a new rapid testing method which works on small- scale websites. • We evaluated that our proposed method works on actual small- scale websites. 30 5. Discussion & Conclusion
  • 31. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Future Works 31 Make a Hypothesis x 2 X Define Metrics f(x) Explore the Solution x⇤ = arg max x2X f(x) We’ve tackled this! Which key metrics we need to focus for effective experiments? How do we define our website as a set of variables? How can we automate the generation of candidates? Website Optimization Process 5. Discussion & Conclusion
  • 32. “Website Optimization Problem and Its Solutions (Paper ID:516)” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo Shuhei Iitsuka, The University of Tokyo. tushuhei.com iitsuka@weblab.t.u-tokyo.ac.jp Thank you for listening. 32
  • 33. “Website Optimization Problem and Its Solutions (Paper ID:516)” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo Appendix 33
  • 34. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. X: Candidate Solutions Y ← {} : Empty Set for Observed Data, n ← 0 : Number of Observations. N_1: Sample Size for Initialization Phase, N_2: Sample Size for Local Search Phase. FOR N_1 TIMES: Y ← Observe(RandomChoice(X)) n++ x* ← LinearAssumption(Y) WHILE n < N DO: x’ ← GetNeighborSolution(x*, X) FOR N_2 TIMES: Y ← Observe(x’) n++ x* ← Update(x*, x’, Y) RETURN x* 34 Initialization Local Search 3. Proposed Testing Method + Streamlined by flexible allocation
  • 35. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. DOE and Linear Assumption • DOE (Design of Experiment) is used in traditional industries
 which have huge cost to reconfigure the environment. • Websites require no cost to change the parameters.
 → We can conduct random observation, then apply ANOVA to estimate each element’s effect. 35 Design of Experiment: Design beforehand. Linear Assumption: Random collection first. Zero
 Reconfiguration Cost 5. Discussion & Conclusion
  • 36. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Webpage Segmentation 36 1ϕ 2ϕ 3ϕ 1 2 ϕ 2 2 ϕ (a) (b)Cai, Deng, et al. Vips: a vision-based page segmentation algorithm. Microsoft technical report, MSR-TR-2003-79, 2003.
  • 37. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Page Element Extraction 37 WELCOME! JOIN NOW Background: {WHITE, BLACK} Button Color: {WHITE, BLACK} Strong Interactive Effect?
  • 38. “Website Optimization Problem and Its Solutions” Shuhei Iitsuka and Matsuo Yutaka, The University of Tokyo. KDD 2015. Bandit Algorithm • ε-greedy • ε: exploration, 1 - ε: exploitation • Softmax: High expected value → High exploitation rate • UCB1: Expected value + Freshness bonus 38 解 x をユーザに表示する確率 p(x) は式 2.1 によって表される。 p(x) = exp(yx/τ) x∈X exp(yx/τ) (2.1) 呼ばれるパラメータであり、探究心の強さを表している。温度が非常に高い わち τ → ∞ のときは解 x を選ぶ確率 p(x) は 1/|X| に収束するため、すべ 等の確率で選ばれることになる。逆に温度が低いときは yx が効き始めるた 価値の期待値が高い解が 1 に近い確率で選ばれるようになる。 CB1 ではこれまでに紹介したアルゴリズムとは異なり、ランダム性を用い 1 では基本的に評価値の期待値 yx が最も高い解を選ぶ戦略だが、解を選ん じてボーナスが追加される。解 x ∈ X を表示した回数を tx とすると、解 x ux は ux = yx + 2 log( x∈X tx) tx 、この UCB 値を最大にする解 x が選択される。 い解を優先的に表示することで実験による損失を免れている。簡単に実 、解の期待値に関わらず探求または活用を選択するため、期待値に大き でも期待値が低い解を選んでしまう可能性がある。 max アルゴリズムでは、解の評価値の期待値に応じて表示する確率を 空間を X、観測データから算出される解 x ∈ X の評価値の期待値を yx をユーザに表示する確率 p(x) は式 2.1 によって表される。 p(x) = exp(yx/τ) x∈X exp(yx/τ) (2.1) れるパラメータであり、探究心の強さを表している。温度が非常に高い τ → ∞ のときは解 x を選ぶ確率 p(x) は 1/|X| に収束するため、すべ 確率で選ばれることになる。逆に温度が低いときは yx が効き始めるた の期待値が高い解が 1 に近い確率で選ばれるようになる。 ではこれまでに紹介したアルゴリズムとは異なり、ランダム性を用い