Data Science of Love
30 likes9,482 views
eHarmony uses a data science approach to matching compatible romantic partners. Its Compatibility Matching System assesses users across 29 Dimensions and also uses Affinity Matching to consider thousands of attributes. Matches are distributed across eHarmony's large network using a graph optimization algorithm that maximizes relationships. This guided communication process aims to help users find long-term relationships successfully.
Data Science of Love
- 2. Data Science of Love
- 3. The eHarmony Difference › Who we are ~45% Tech
- 4. The eHarmony Difference › Who we are ~45% Tech ~15% Customer Care
- 5. The eHarmony Difference › Who we are ~45% Tech ~15% Customer Care ~10% Marketing
- 6. The eHarmony Difference › Compatibility Matching System®
- 7. The eHarmony Difference › Compatibility Matching System® Compatibility Matching System®
- 8. The eHarmony Difference › Compatibility Matching System® Compatibility Matching System® 1 Compatibility Matching
- 9. The eHarmony Difference › Compatibility Matching System® Compatibility Matching System® 1 2 Compatibility Affinity Matching Matching
- 10. The eHarmony Difference › Compatibility Matching System® Compatibility Matching System® 1 2 3 Compatibility Affinity Match Matching Matching Distribution
- 12. The eHarmony Difference › Compatibility Matching System® 1 2 3 Compatibility Affinity Match Matching Matching Distribution
- 13. The eHarmony Difference › Compatibility Matching System® 1 2 3 Compatibility Affinity Match Matching Matching Distribution
- 14. Compatibility Matching › Relationship Questionnaire › Personality › Values › Attitudes › Beliefs
- 15. Compatibility Matching › Obstreperousness
- 16. Compatibility Matching › Romantic
- 17. Compatibility Matching › 29 Dimensions®
- 18. Compatibility Matching › 29 Dimensions®
- 19. The eHarmony Difference › Compatibility Matching System® 1 2 3 Compatibility Affinity Match Matching Matching Distribution
- 20. The eHarmony Difference › Compatibility Matching System® Layers on Top of Compatibility Matching 1 2 3 Compatibility Affinity Match Matching Matching Distribution
- 22. Affinity Matching › 61 21
- 23. Affinity Matching › 61 21 3000
- 24. Affinity Matching › 61 21 3000
- 26. Affinity Matching › … … …
- 27. Affinity Matching › Distance Prob( )
- 28. Affinity Matching › Distance
- 29. Affinity Matching › Height difference Prob( ) 4 -‐ 8 in cm
- 30. Affinity Matching › “Attractiveness” Prob( )
- 31. Affinity Matching › Zoom level
- 32. Affinity Matching › Zoom level
- 33. Affinity Matching › Zoom level
- 34. Affinity Matching › Food preference 25% -‐1% -‐24% 20% 13% 9% -‐5% -‐27% 7% 0% -‐12% -‐21% -‐42% -‐19% -‐23% 19% 0% -‐28% 28% 10% 9% -‐11% -‐35% 11% 44%
- 35. Affinity Matching › Food preference 25% -‐1% -‐24% 20% 13% 9% -‐5% -‐27% 7% 0% -‐12% -‐21% -‐42% -‐19% -‐23% 19% 0% -‐28% 28% 10% 9% -‐11% -‐35% 11% 44%
- 36. Affinity Matching › Prob( | data) ~40M registered users ... ... ~10^3 aCributes ? ~10^8 matches per day ~10^8 daily
- 37. Affinity Matching › Prob( | data) data large sparse redundant dense complex non-‐linear (freetext) moderate #features vowpal wabbit gradient boosted decision trees ?
- 38. Affinity Matching › Domain Specific Language ... ... ... ?? DSL embarassingly parallel
- 39. The eHarmony Difference › Compatibility Matching System® 1 2 3 Compatibility Affinity Match Matching Matching Distribution
- 40. The eHarmony Difference › Compatibility Matching System® Delivering the right matches at the right time to as many people as possible across the entire network. 1 2 3 Compatibility Affinity Match Matching Matching Distribution
- 41. Match Distribution › Graph optimization
- 42. Match Distribution › Graph optimization
- 43. Match Distribution › Graph optimization 2 2
- 44. Match Distribution › Graph optimization 2 1 2
- 45. Match Distribution › Graph optimization 2 Prob( 1 | data) 2
- 46. Match Distribution › Graph optimization 2 Prob( 1 | data) 2
- 47. Match Distribution › Graph optimization 2 Prob( | data) 2
- 48. Match Distribution › Graph optimization 2 Prob( | data) 2
- 49. Guided Resulting Customer Experience › Communication
- 50. Guided Resulting Customer Experience › Communication
- 51. Guided Resulting Customer Experience › Communication ? !
- 52. Resulting Customer Experience › Success!
- 53. Resulting Customer Experience › Success!
- 54. eHarmony Results › The eHarmony Impact 2005 90 eHarmony Members Married Every Day * * According to a Harris Interactive survey for eHarmony
- 55. eHarmony Results › The eHarmony Impact 2005 2007 236 eHarmony Members Married Every Day * * According to a Harris Interactive survey for eHarmony
- 56. eHarmony Results › The eHarmony Impact 2005 2007 2009 542 eHarmony Members Married Every Day * * According to a Harris Interactive survey for eHarmony
- 57. @petricek linkedin.com/in/petricek