10 Pitfalls in Data Science - LA Data Science Meetup Kick-Off - Feb 2014

Ce8e94cc306ba164175f693fb01aa8b0?s=47 szilard
March 18, 2014
140

10 Pitfalls in Data Science - LA Data Science Meetup Kick-Off - Feb 2014

Ce8e94cc306ba164175f693fb01aa8b0?s=128

szilard

March 18, 2014
Tweet

Transcript

  1. 10 Pitfalls in Data Science Szilárd Pafka, PhD Chief Scientist,

    Epoch LA Machine Learning Meetup Data Science Track Feb 2014
  2. About me

  3. Data Science

  4. (Some) Pitfalls • DS = IT project • DS isolated

    from business • Restricted access to data • Not enough EDA/cleaning • Data leakage • Overfitting • Optimizing wrong metric • Skip model validation • Too complex to deploy • Poor communication
  5. Contact [email removed from slideshare] www.linkedin.com/in/szilard