Great article! I like the way of you introduce these paradoxes, and the example provided is very clear and easy to understand. This article makes me wondering how are we going to distinguish the latent variables, collinear variables and class imbalance when we are doing a statistical study, and how should we avoid these types of paradoxes. If a paradox appears, does this mean the dataset we used was biased.

Great article! As a statistic student I am very familiar with the features such as mean or median, and learned to use distributions to analyze the probability. However, I never learned about the reduction of dimensionality. I was wondering how we could related the correlations of different variables as dimensions?