/sci/ - Science & Math » Thread #11667126

33KiB, 692x692, Blog_pca_6b.png

View Same Google iqdb SauceNAO

Anonymous Wed 13 May 11:58:52 2020 No.11667126 View Reply Original Report

Quoted By: >>11667130 >>11667220 >>11667268 >>11668777 >>11669720

Guys I'm taking a course in Principal Component Analysis (PCA) and I don't fucking understand shit

Anonymous

Anonymous Wed 13 May 2020 12:01:22 No.11667130 Report

Quoted By:

>>11667126
What is exactly you don’t understand? Its just a coordinate change

Anonymous

Anonymous Wed 13 May 2020 12:29:56 No.11667184 Report

Quoted By: >>11667330

theres a whole course on pca?

Anonymous

Anonymous Wed 13 May 2020 12:43:25 No.11667220 Report

Quoted By:

>>11667126
>course in Principal Component Analysis
>course
Reminds me of the statistics courses that civil engineers have to take at my university, where they do averages and variance for a whole semester.

I suggest getting some material on covariance and then look for an online course in coursera or edx or so that teaches statistics. One of the lessons will be PCA and you'll understand.

Anonymous

Anonymous Wed 13 May 2020 13:09:48 No.11667268 Report

Quoted By:

>>11667126
You are basically taking a set of data that is multidimensional and squishing it onto a lower dimension so that you have less dimensions to work with. The entire thing is based upon reducing the loss in data from the dimensionality reduction

Anonymous

Anonymous Wed 13 May 2020 13:14:56 No.11667278 Report

Quoted By: >>11667299

Isn't PCA just a particular case of SVD decomposition and eigen-decomposition?

Anonymous

Anonymous Wed 13 May 2020 13:24:50 No.11667299 Report

Quoted By:

>>11667278
yes it is, you just work with a square matrix instead of a rectangular. When you use the more general case its called LDA (Linear Discriminant Analysis)

Anonymous

Anonymous Wed 13 May 2020 13:35:27 No.11667330 Report

Quoted By:

>>11667184
Lol was thinking the same

Anonymous

Anonymous Wed 13 May 2020 21:52:16 No.11668777 Report

Quoted By:

>>11667126
its literally just glorified linear regression. What's so hard about it lol

Anonymous

Anonymous Thu 14 May 2020 03:14:14 No.11669720 Report

Quoted By:

>>11667126
You have a bunch of data with a billion dimensions, but actually most of the information is contained in like six of them, so you get rid of the other nine hundred and ninetynine million nine hundred and ninety-nine thousand nine hundred and four. In Matlab it takes two lines of code. I can’t believe they would teach an entire course on PCA.

Anonymous

Anonymous Thu 14 May 2020 03:29:11 No.11669767 Report

Quoted By:

So being engineer is all fun and games until you need to use or understand real math but you cheated on calculus and actually never read any proof because reading is for losers.

Anonymous

Anonymous Thu 14 May 2020 03:34:14 No.11669781 Report

Quoted By:

But the again it may happen if you are a mathematician.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches