4 ms·
Do you really understand Principal Component Analysis?
- catnaroek 9y agoAn “explanation” of principal component analysis that somehow manages not to mention the geometric meaning of the covariance matrix and its eigenvectors? Sad. Also, please learn to use LaTeX correctly. Don't use the same font for English text and formulas.
- newen 9y agoNothing about how it relates to the multivariate Gaussian distribution either.
- aptrishu 9y agoEverything can't be covered in a single blog post. As it is mentioned in the post, it will be a series of posts. Thanks for your review. :)
- aptrishu 9y agoHi, the main focus of the post was not to totally focus on geometrical interpretation but rather on why we use eigenvectors and all. I could have made it simpler, had I chosen to represent all of them geometrically. Thanks for your review, I could have done better.
- catnaroek 9y agoPresumably your target audience cares about the application of PCA, rather than the mathematical technique for its own sake. Thus, you need to answer the question “What does it do for me?” before the question “How does it work?” I believe that the answers to these two questions are, respectively, the geometric interpretation of a PCA, and the algorithm for computing a SVD decomposition. But I could be wrong! If you think I'm wrong, please tell me how.
- aptrishu 9y agoAlso, medium doesn't support latex. I could have resized the exported images properly but it was too much of a trouble doing that for every image.
- catnaroek 9y agoIt's not about whether Medium supports LaTeX. Just look at this: https://cdn-images-1.medium.com/max/1600/1*APPNIwQqDG4PVGyLLQ32FQ.png https://cdn-images-1.medium.com/max/1600/1*APPNIwQqDG4PVGyLL...