Latest / Logosofia no Dia a Dia
Correlation Beyond Pearson: 5 Techniques You Need
Most people learn Pearson's r in week one of statistics and then get it wrong for the rest of their career. This episode breaks down the full spectrum of correlation analysis — from the classic Pearson, Spearman, and Kendall measures through partial correlation for controlling confounders, distance correlation for detecting any dependence at all, and canonical correlation analysis for multi-variable sets. We cover the pitfalls of outliers, the lurking variable problem, and why you should always plot your data before reporting a coefficient. Whether you're analyzing ice cream sales and…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.