Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

From a sea of projects to collaboration opportunities within seconds

83 views

Published on

Using multifactor-analysis to efficiently analyse large amounts of data in order to find a cluster of closely related projects for business opportunities, partnerships, competitor analysis and collaboration

Published in: Data & Analytics
  • Be the first to comment

  • Be the first to like this

From a sea of projects to collaboration opportunities within seconds

  1. 1. From a sea of projects to collaboration opportunities within seconds Michel Drescher, Cloud computing standards specialist OeRC, University of Oxford, UK
  2. 2. This is your project. 2
  3. 3. This is your project among H2020. 3 8400+ projects currently funded under H2020. Source: European Commission You’ve got a problem… Hello, project! How will you ever find potential collaboration projects effectively and efficiently?
  4. 4. 4 That is impossible.
  5. 5. CloudWATCH has had an idea… 5
  6. 6. The 1-slide explanation 6 38 responses, scoring the importance of NIST Cloud characteristics for them selves Out of these, your collaboration opportunities lie within this cluster! (Had you provided your scores…)
  7. 7. 7 This is how it works.
  8. 8. 1. Data taking / scoring “How important is … for your work/service/project?” For 13 Cloud computing characteristics 8
  9. 9. 1. Data taking / scoring 9Essential Common NIST SP 800-145
  10. 10. 2. Interactive analysis 10
  11. 11. 3. Access underlying data 11 2-step process: • Submit scores • Offline verification
  12. 12. A bit of statistics Principal Component Analysis (Karl Pierson 1901, Harold Hotelling 1930) Multi-variance analysis Emphasises variance and patterns in data Data transformation for dimension reduction in analysis Hierarchical clustering using Euclidian distance Form clusters of “similar” respondents, starting with 1-member clusters “Similarity” (i.e., distance) calculated using Euclidian distance function Distance between clusters uses “weighed pair-group centroid”  performs well with large variance in cluster sizes 12
  13. 13. 13 Can I have a go?
  14. 14. Yes you can. But first, you must submit your scores! SESA – 10 of 19 projects responded ICEI – 4 of 13 projects responded NATRES – 11 of 20 projects responded DPSP – 3 of 24 projects responded 14 https://tethys.oerc.ox.ac.uk:8443/cluster/index.xhtml
  15. 15. 15 Demo time! (if agenda permits…)

×