Correlation vs. cosine similarity

RelatedInsightsHighlights

Correlation Doesn't Mean Causation Just because people with long toenails watch lots of TV doesn't mean watching TV causes long toenails - it's just a connection. The same applies to fat consumption and heart disease in these early studies.

Causation vs Correlation 📜

Causation: one event directly produces another.

Correlation: Two or more events occur together, but it is not proven that one caused the other. (Confirmation bias will want you to believe there is a connection).

First thing we'll do: highlight the two *very* different motivations for linear models

1) for causal inference

2) for prediction

Karl Rohe • Tweet