Predictive Modeling & Data-Driven Product Insights atLinkedIn
This talk: the two sides of data at LinkedInUsing data to build products that delight our usersUsing data to uncover insig...
Connect the world’sprofessionals to make themmore productive andsuccessful
150M+ professional profiles
What can we do with all of this data?Build products.
150MM+ professional profiles
Data tools and infrastructure
Data science
What can we do with all of this data?Derive insights that are actionable andimprove the business or our members’experience...
Panel Data EconometricsIdentifying site activities that predict future engagement
What can we do with all of this data?Derive insights that are just plain cool.
What can we do with all of this data?Insights lead to products.And what can big data products do?
We are good at getting people to make         different decisions……but we can do more to help people make           better...
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous
Upcoming SlideShare
Loading in...5
×

Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous

600

Published on

Talk given at Advanced Analytics & Big Data Forum conference in San Francisco on April 25, 2012.

Abstract: Data on 150+ million professionals' careers and networks provide a fascinating playground for analysts to discover data insights about career trends, the social web and the economy. This talk will focus on how insights extracted from the LinkedIn dataset enable individuals with limited information the ability to make better decisions about their professional lives. In the course of this theme we will discuss data tools, insights and approaches to predictive modeling in the context of the LinkedIn dataset and Analytics Team.

Published in: Technology, Education
1 Comment
3 Likes
Statistics
Notes
No Downloads
Views
Total Views
600
On Slideshare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
0
Comments
1
Likes
3
Embeds 0
No embeds

No notes for slide
  • Some stats4BN searches150M members60M USMembers first, monetization comes second
  • 2 members a second…Plus dynamics over time.Just an awesome datasetEven though only been around since 2003, we have data going much further back because our members' careers span that timeGoal: build simple, brilliant products that delight our users. ME: and use data to enhance those products where applicable
  • Two thingsBuild data productsData insights…going to talk a bit more about that in the second part of the talk
  • How many of these products are data driven? All of them.
  • How many of these products are data driven? All of them.
  • 2 members a second…Plus dynamics over time.Just an awesome datasetEven though only been around since 2003, we have data going much further back because our members' careers span that timeGoal: build simple, brilliant products that delight our users. ME: and use data to enhance those products where applicable
  • How many of these products are data driven? All of them.
  • Two thingsBuild data productsData insights…going to talk a bit more about that in the second part of the talk
  • Over 75TB/day processedOver 10BN rows / dayReal time availability for key eventsMost tracking events available after 15 minutes via kafka and hadoop
  • Two thingsBuild data productsData insights…going to talk a bit more about that in the second part of the talk
  • Ultimately it’s not about data or tools, it’s about asking the right questions and employing star data scientists who own the end to end. Examples of how we work…
  • Ultimately it’s not about data or tools, it’s about asking the right questions and employing star data scientists who own the end to end. Examples of how we work…
  • Two thingsBuild data productsData insights…going to talk a bit more about that in the second part of the talk
  • Panel data: Following observations over time allows us to control for subject-specific (unobservable) effects Going further away from the gold standard of A/B testing and moving closer to establishing predictive power
  • Two thingsBuild data productsData insights…going to talk a bit more about that in the second part of the talk
  • Look at the length of the names – now that’s an interesting story! There’s Chip, Todd and Trey - the quintessential sales guys. CEOs are more diverse – but they still want to be your friend -- so they use nicknames.
  • Look at the length of the names – now that’s an interesting story! There’s Chip, Todd and Trey - the quintessential sales guys. CEOs are more diverse – but they still want to be your friend -- so they use nicknames.
  • Which companies are over-represented in founders’ histories?
  • Which companies are over-represented in founders’ histories?
  • Two thingsBuild data productsData insights…going to talk a bit more about that in the second part of the talk
  • Two thingsBuild data productsData insights…going to talk a bit more about that in the second part of the talk
  • Transcript of "Predictive Modeling & Data-Driven Product Insights at LinkedIn - Scott Nicholson / @scootrous"

    1. 1. Predictive Modeling & Data-Driven Product Insights atLinkedIn
    2. 2. This talk: the two sides of data at LinkedInUsing data to build products that delight our usersUsing data to uncover insights
    3. 3. Connect the world’sprofessionals to make themmore productive andsuccessful
    4. 4. 150M+ professional profiles
    5. 5. What can we do with all of this data?Build products.
    6. 6. 150MM+ professional profiles
    7. 7. Data tools and infrastructure
    8. 8. Data science
    9. 9. What can we do with all of this data?Derive insights that are actionable andimprove the business or our members’experience.Question: What actions on the site arepredictive of future engagement?
    10. 10. Panel Data EconometricsIdentifying site activities that predict future engagement
    11. 11. What can we do with all of this data?Derive insights that are just plain cool.
    12. 12. What can we do with all of this data?Insights lead to products.And what can big data products do?
    13. 13. We are good at getting people to make different decisions……but we can do more to help people make better decisions.

    ×