SlideShare une entreprise Scribd logo
1  sur  18
Télécharger pour lire hors ligne
Clustering Using
Integer Programming
Katie Kuksenok
Michael Brooks
What are we doing?
• Investigating IP formulations of the clustering
  problem

• Developing a web application for clustering

• Applying clustering to several datasets

• Visualizing resulting clusters
The Clustering Problem
• Input:
  ▫ N entities
  ▫ Each entity has a value for each of M features

• Goal:
  ▫ Partition the entities “optimally”

• Issues with clustering:
  ▫ Definition of “optimal”
  ▫ Different feature types don’t work well together
The Clustering Problem
Exposes high-level properties of data

Important for exploratory data analysis in data
 mining

Finding the optimal clustering is NP-Complete
Grotschel and Wakabayashi IP
• Recall the Grotschel-Wakabayashi formulation
  (clustering wild cats):

• Variables: xij = 1 if i and j are in the same cluster

• Data: wij = disagreements(i,j) – agreements(i,j)

• Minimize: SUM { wij xij : over all i, j }
Formulation: Constraints
• Constraints for reflexivity, symmetry, transitivity
  reduce to the following constraints:
Exploratory Clustering Application
• Currently in development

• User provides data to website
• Web server processes data
  ▫ Produces OPL code
  ▫ Runs the model in LPSolve (another MIP solver)
• Results are returned to the user, with visuals
Preliminary Exploration
• With working parts of application:
• can cluster and visualize datasets

• verified Grotschel/Wakabayashi clustering results
• used clustering to explore senate voting records
Data Sets (Cats)
• Cluster 30 varieties of wild cats (from paper)
• Cats have 14 features
• Reproduced results from paper

• Model has 12180 constraints and 435 variables
• Runs in less than 1 second
• No branching
Data Sets (Senate Voting Records)
• 100 senators
• Senate records for 8 passed bills from this year

• Model has 485100 constraints and 4950 variables
• Runs in 1 minute
• No branching
Data Sets (Senate Voting By State)
• Grouped senate voting by state
• If senators for a state disagreed, ‘Maybe’

• Model has 58800 constraints and 1225 variables
• Runs in 1.5 seconds
• No branching
What now?
• The Application
 ▫ We’d like to put it all together
• Branch and Bound
 ▫ No real-life datasets so far have required branching
 ▫ We have constructed datasets that did
 ▫ Can we determine under what conditions branching
   will be required?
This is a guepard (cheetah in French?):




       …any other questions?
Approaches to Clustering

             • What makes a good clustering
               algorithm?
              ▫ Runtime and space efficiency (when N is
                large)
     Many
algorithms
              ▫ Ability to quickly incorporate new data
              ▫ Ability to deal with many-dimensional
        IP      data (when M is large)
              ▫ Optimality (closeness to)

Contenu connexe

En vedette

Storyboard(printscreen)
Storyboard(printscreen)Storyboard(printscreen)
Storyboard(printscreen)
Sam Arblaster
 

En vedette (8)

Helping Computers Find Meaning they Lost in Translation
Helping Computers Find Meaning they Lost in TranslationHelping Computers Find Meaning they Lost in Translation
Helping Computers Find Meaning they Lost in Translation
 
Games on Social Networks: Balancing Friendships
Games on Social Networks: Balancing FriendshipsGames on Social Networks: Balancing Friendships
Games on Social Networks: Balancing Friendships
 
Storyboard(printscreen)
Storyboard(printscreen)Storyboard(printscreen)
Storyboard(printscreen)
 
Uncertainty in Chronic Illness and Patients' Online Experience
Uncertainty in Chronic Illness and Patients' Online ExperienceUncertainty in Chronic Illness and Patients' Online Experience
Uncertainty in Chronic Illness and Patients' Online Experience
 
Michel Foucault
Michel FoucaultMichel Foucault
Michel Foucault
 
Brand management Full notes
Brand management Full notesBrand management Full notes
Brand management Full notes
 
Mishel's Uncertainty in Illness Theory
Mishel's Uncertainty in Illness TheoryMishel's Uncertainty in Illness Theory
Mishel's Uncertainty in Illness Theory
 
Patricia Benner (Novice to Expert Theory)
Patricia Benner (Novice to Expert Theory)Patricia Benner (Novice to Expert Theory)
Patricia Benner (Novice to Expert Theory)
 

Similaire à IP Clustering as Exploratory Data Analysis

Computer Vision for Beginners
Computer Vision for BeginnersComputer Vision for Beginners
Computer Vision for Beginners
Sanghamitra Deb
 
H2O World - Intro to Data Science with Erin Ledell
H2O World - Intro to Data Science with Erin LedellH2O World - Intro to Data Science with Erin Ledell
H2O World - Intro to Data Science with Erin Ledell
Sri Ambati
 

Similaire à IP Clustering as Exploratory Data Analysis (20)

Entity embeddings for categorical data
Entity embeddings for categorical dataEntity embeddings for categorical data
Entity embeddings for categorical data
 
BitBootCamp Evening Classes
BitBootCamp Evening ClassesBitBootCamp Evening Classes
BitBootCamp Evening Classes
 
Machine Learning With ML.NET
Machine Learning With ML.NETMachine Learning With ML.NET
Machine Learning With ML.NET
 
InfoEducatie - Face Recognition Architecture
InfoEducatie - Face Recognition ArchitectureInfoEducatie - Face Recognition Architecture
InfoEducatie - Face Recognition Architecture
 
Scaling Face Recognition with Big Data - Key Notes at DevTalks Bucharest 2017
Scaling Face Recognition with Big Data - Key Notes at DevTalks Bucharest 2017Scaling Face Recognition with Big Data - Key Notes at DevTalks Bucharest 2017
Scaling Face Recognition with Big Data - Key Notes at DevTalks Bucharest 2017
 
Lessons from lhc
Lessons from lhcLessons from lhc
Lessons from lhc
 
Master guide to become a data scientist
Master guide to become a data scientist Master guide to become a data scientist
Master guide to become a data scientist
 
Master guide to become a Data Scientist -by zekeLabs
Master guide to become a Data Scientist  -by zekeLabsMaster guide to become a Data Scientist  -by zekeLabs
Master guide to become a Data Scientist -by zekeLabs
 
[RSS2023] Local Object Crop Collision Network for Efficient Simulation
[RSS2023] Local Object Crop Collision Network for Efficient Simulation[RSS2023] Local Object Crop Collision Network for Efficient Simulation
[RSS2023] Local Object Crop Collision Network for Efficient Simulation
 
Research network infrastructure engineers
Research network infrastructure engineersResearch network infrastructure engineers
Research network infrastructure engineers
 
Computer Vision for Beginners
Computer Vision for BeginnersComputer Vision for Beginners
Computer Vision for Beginners
 
Machine Learning Essentials Demystified part1 | Big Data Demystified
Machine Learning Essentials Demystified part1 | Big Data DemystifiedMachine Learning Essentials Demystified part1 | Big Data Demystified
Machine Learning Essentials Demystified part1 | Big Data Demystified
 
Zestimate Lambda Architecture
Zestimate Lambda ArchitectureZestimate Lambda Architecture
Zestimate Lambda Architecture
 
SMART International Symposium for Next Generation Infrastructure: The roles o...
SMART International Symposium for Next Generation Infrastructure: The roles o...SMART International Symposium for Next Generation Infrastructure: The roles o...
SMART International Symposium for Next Generation Infrastructure: The roles o...
 
How we integrate Machine Learning Algorithms into our IT Platform at Outfittery
How we integrate Machine Learning Algorithms into our IT Platform at OutfitteryHow we integrate Machine Learning Algorithms into our IT Platform at Outfittery
How we integrate Machine Learning Algorithms into our IT Platform at Outfittery
 
Webinar: SQL for Machine Data?
Webinar: SQL for Machine Data?Webinar: SQL for Machine Data?
Webinar: SQL for Machine Data?
 
H2O World - Intro to Data Science with Erin Ledell
H2O World - Intro to Data Science with Erin LedellH2O World - Intro to Data Science with Erin Ledell
H2O World - Intro to Data Science with Erin Ledell
 
A data analyst view of Bigdata
A data analyst view of Bigdata A data analyst view of Bigdata
A data analyst view of Bigdata
 
Counting Unique Users in Real-Time: Here's a Challenge for You!
Counting Unique Users in Real-Time: Here's a Challenge for You!Counting Unique Users in Real-Time: Here's a Challenge for You!
Counting Unique Users in Real-Time: Here's a Challenge for You!
 
Big Data
Big DataBig Data
Big Data
 

Dernier

The basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxThe basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptx
heathfieldcps1
 

Dernier (20)

Application orientated numerical on hev.ppt
Application orientated numerical on hev.pptApplication orientated numerical on hev.ppt
Application orientated numerical on hev.ppt
 
Nutritional Needs Presentation - HLTH 104
Nutritional Needs Presentation - HLTH 104Nutritional Needs Presentation - HLTH 104
Nutritional Needs Presentation - HLTH 104
 
PROCESS RECORDING FORMAT.docx
PROCESS      RECORDING        FORMAT.docxPROCESS      RECORDING        FORMAT.docx
PROCESS RECORDING FORMAT.docx
 
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
 
fourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writingfourth grading exam for kindergarten in writing
fourth grading exam for kindergarten in writing
 
The basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxThe basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptx
 
Holdier Curriculum Vitae (April 2024).pdf
Holdier Curriculum Vitae (April 2024).pdfHoldier Curriculum Vitae (April 2024).pdf
Holdier Curriculum Vitae (April 2024).pdf
 
Mattingly "AI & Prompt Design: The Basics of Prompt Design"
Mattingly "AI & Prompt Design: The Basics of Prompt Design"Mattingly "AI & Prompt Design: The Basics of Prompt Design"
Mattingly "AI & Prompt Design: The Basics of Prompt Design"
 
Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"
Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"
Mattingly "AI & Prompt Design: Structured Data, Assistants, & RAG"
 
Sports & Fitness Value Added Course FY..
Sports & Fitness Value Added Course FY..Sports & Fitness Value Added Course FY..
Sports & Fitness Value Added Course FY..
 
APM Welcome, APM North West Network Conference, Synergies Across Sectors
APM Welcome, APM North West Network Conference, Synergies Across SectorsAPM Welcome, APM North West Network Conference, Synergies Across Sectors
APM Welcome, APM North West Network Conference, Synergies Across Sectors
 
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17  How to Extend Models Using Mixin ClassesMixin Classes in Odoo 17  How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
 
Unit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptxUnit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptx
 
Z Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot GraphZ Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot Graph
 
Unit-V; Pricing (Pharma Marketing Management).pptx
Unit-V; Pricing (Pharma Marketing Management).pptxUnit-V; Pricing (Pharma Marketing Management).pptx
Unit-V; Pricing (Pharma Marketing Management).pptx
 
Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17Advanced Views - Calendar View in Odoo 17
Advanced Views - Calendar View in Odoo 17
 
Grant Readiness 101 TechSoup and Remy Consulting
Grant Readiness 101 TechSoup and Remy ConsultingGrant Readiness 101 TechSoup and Remy Consulting
Grant Readiness 101 TechSoup and Remy Consulting
 
microwave assisted reaction. General introduction
microwave assisted reaction. General introductionmicrowave assisted reaction. General introduction
microwave assisted reaction. General introduction
 
psychiatric nursing HISTORY COLLECTION .docx
psychiatric  nursing HISTORY  COLLECTION  .docxpsychiatric  nursing HISTORY  COLLECTION  .docx
psychiatric nursing HISTORY COLLECTION .docx
 
Introduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsIntroduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The Basics
 

IP Clustering as Exploratory Data Analysis

  • 2. What are we doing? • Investigating IP formulations of the clustering problem • Developing a web application for clustering • Applying clustering to several datasets • Visualizing resulting clusters
  • 3. The Clustering Problem • Input: ▫ N entities ▫ Each entity has a value for each of M features • Goal: ▫ Partition the entities “optimally” • Issues with clustering: ▫ Definition of “optimal” ▫ Different feature types don’t work well together
  • 4. The Clustering Problem Exposes high-level properties of data Important for exploratory data analysis in data mining Finding the optimal clustering is NP-Complete
  • 5. Grotschel and Wakabayashi IP • Recall the Grotschel-Wakabayashi formulation (clustering wild cats): • Variables: xij = 1 if i and j are in the same cluster • Data: wij = disagreements(i,j) – agreements(i,j) • Minimize: SUM { wij xij : over all i, j }
  • 6. Formulation: Constraints • Constraints for reflexivity, symmetry, transitivity reduce to the following constraints:
  • 7. Exploratory Clustering Application • Currently in development • User provides data to website • Web server processes data ▫ Produces OPL code ▫ Runs the model in LPSolve (another MIP solver) • Results are returned to the user, with visuals
  • 8. Preliminary Exploration • With working parts of application: • can cluster and visualize datasets • verified Grotschel/Wakabayashi clustering results • used clustering to explore senate voting records
  • 9. Data Sets (Cats) • Cluster 30 varieties of wild cats (from paper) • Cats have 14 features • Reproduced results from paper • Model has 12180 constraints and 435 variables • Runs in less than 1 second • No branching
  • 10.
  • 11. Data Sets (Senate Voting Records) • 100 senators • Senate records for 8 passed bills from this year • Model has 485100 constraints and 4950 variables • Runs in 1 minute • No branching
  • 12.
  • 13. Data Sets (Senate Voting By State) • Grouped senate voting by state • If senators for a state disagreed, ‘Maybe’ • Model has 58800 constraints and 1225 variables • Runs in 1.5 seconds • No branching
  • 14.
  • 15. What now? • The Application ▫ We’d like to put it all together • Branch and Bound ▫ No real-life datasets so far have required branching ▫ We have constructed datasets that did ▫ Can we determine under what conditions branching will be required?
  • 16. This is a guepard (cheetah in French?): …any other questions?
  • 17.
  • 18. Approaches to Clustering • What makes a good clustering algorithm? ▫ Runtime and space efficiency (when N is large) Many algorithms ▫ Ability to quickly incorporate new data ▫ Ability to deal with many-dimensional IP data (when M is large) ▫ Optimality (closeness to)