Wednesday, September 18, 2013

How to consume statistical analysis

I am giving a talk next week for the Data Science Group in Cambridge.  It's part six of the Leading Analytics series:

Building your Analytical Skill set


  • Tuesday, September 24, 2013
    6:00 PM to 
  • 1 Memorial Dr, CambridgeMA (map)
    Commons
  • Price: $10.00/per person
The subtitle is "How to be a good consumer of statistical analysis."  My goal is to present (in about 70 minutes) some basic statistical knowledge a manager should have to work with an analysis team.

Part of the talk is about how to interact with the team: I will talk about an exploration process that is collaborative between analysts and managers, and iterative.

And then I'll introduce topics in statistics, including lots of material from this blog:

  • The CDF: the best, and sadly underused, way to visualize distributions.
  • Scatterplots, correlation and regression: how to visualize and quantify relationships between variables.
  • Hypothesis testing: the most abused tool in statistics.
  • Estimation: quantifying and working with uncertainty.
  • Visualization: how to use the most powerful and versatile data analysis system in the world, human vision.
Here are the slides I'm planning to present:



Hope you can attend!

5 comments:

  1. Friggin' brilliant. I'm making this a required reading in my basic statistics course RIGHT NOW. I'd give my students extra credit for attending the talk, but it's a long hike from San Antonio, Texas.

    ReplyDelete
    Replies
    1. Thanks! I'm sorry you and your students won't be able to make it to the meeting :)

      Delete
  2. Great slides but I think it's tough to properly understand p-values without understanding power. Interpretation of a p-value depends on power (http://bayesianbiologist.com/2013/10/17/p-value-fallacy-alive-and-well-latest-case-in-the-journal-nature/). I might be over-thinking it, but in my experience it's this complexity that leads to misinterpretation. People (myself included) want to interpret p-values as a non-conditional probability, often the probability that the null is not true, but that's incorrect.

    By the way, thank you very very much for making your books freely available. They're a great resource.

    ReplyDelete
    Replies
    1. That's a good point -- thanks for this comment and for the link.

      Delete