Proposal for new stat pages

[TommasoGhilardi]

Here are some ideas I’d like to turn into new pages:

• Outliers or wrong distribution — a supplementary page where we talk about:

  • spotting outliers in data
  • spotting outliers in models
  • asking whether they’re really outliers, or if we’re just using the wrong distribution. (I’d bring two examples: lognormal and Student’s t.)

• Model assumptions — maybe not a full page, but I’d like to expand on the point that data itself doesn’t have to be normal. What matters is whether the residuals are normally distributed.

These pages could live under a new subsection like “Dig Deeper” or “Extra Insights.”