Nice post by the way, I’d comment that the axes (the angles) look a little bit strange to represent the 3d plane in some moment I thought wait why 3D vectors if the draw are in 2D space? Then I saw again and looked third dimension.
I always look at this with the glasses of causality. Let's assume we have 3 features like A (Altitude), T (Temperature) and AQ (Air Quality). We have 10 data points of this 3 variables, where A is the root-cause of T and AD, while T and AQ are purely correlated. How would the Gram-Schmidt process work in this example? Would it reveal the correlation from the causality?
Epsilons, no. 4: The Gram-Schmidt process
Nice post by the way, I’d comment that the axes (the angles) look a little bit strange to represent the 3d plane in some moment I thought wait why 3D vectors if the draw are in 2D space? Then I saw again and looked third dimension.
There is a mistake in the plot called final step, it’s the orthogonal projection of a3. The formulas are ok, but not the text
Thank you for presenting your great work!
I always look at this with the glasses of causality. Let's assume we have 3 features like A (Altitude), T (Temperature) and AQ (Air Quality). We have 10 data points of this 3 variables, where A is the root-cause of T and AD, while T and AQ are purely correlated. How would the Gram-Schmidt process work in this example? Would it reveal the correlation from the causality?
Nice job! One question I have never asked to you, how do you create such great illustrations?
Good explanation 👑