Principal Component Analysis (PCA)
Video Overview & Insights
This video is gentle and motivated introduction to Principal Component Analysis (PCA). We use PCA to analyze the 2021 World Happiness Report published 2021 and discover what makes countries truly happy. :)
Ur english accent sounds moroccan
References:
- Scikit-Learn User Guide : https://scikit-learn.org/stable/modules/decomposition.html#pca
Please use correct map of India π
- A Tutorial on Principal Component Analysis: https://arxiv.org/abs/1404.1100
- Andrew Ng Stanford Course: https://www.youtube.com/watch?v=ey2PE5xi9-A#t=2385&ab_channel=Stanford
Im gonna reply to my comment when I finally understand
- Kaggle dataset: https://www.kaggle.com/ajaypalsinghlo/world-happiness-report-2021
--------------------------
Addictive and satisfying visuals
Timestamps:
0:00 Intro
thank you !
1:37 Projecting a point on a line
2:00 Optimization
Excellent!
3:27 First component
4:19 Second component
0:45 I don't think this is correct visualization: PCA is a pure rotation of vector basis - why do you start from non orthogonal Social, Life, GDP vector basis and then rectify it into orthogonal PCA features? PCA doesn't change angles between data points.
5:20 More generally ...
--------------------------
wow!!!
Credit:
π Manim and Python : https://github.com/3b1b/manim
What do the numbers (1.1, -0.9 and so on) represent? It does not make sense without knowing that.
π΅ Blender3D: https://www.blender.org/
ποΈ Emacs: https://www.gnu.org/software/emacs/
πΉ Intro Music: Waltz of the Flowers - Tchaikovsky
πΉ Outro Music: Like That - Anno Domini Beats
Type of things I want to see π²π¦ ranking high at.
Allah i yesser lik!
This video would not have been possible without the help of Gâkçe Dayanıklı.
Interesting overlap between economics and statistics!
More User Perspectives
You explained this so well! Thank you!!!
@cristinabarros1719Amazing vid, thank you very much for highlighting the important intuitions!
@markshenouda5456Are the magnitudes of the components correlated to the magnitude of the target variable, i.e. happiness? If so, how does it make sense that the second component is balance? Balance implies that you want a score as close to zero as possible, i.e. individual factors being equal to social. But if we're trying to maximise the output of the components in order to maximise happiness, how does a score closer to zero make sense?
Having a look at the data sources it seems like "Social" is a measure of social support, corruption, etc, while "GDP" is log GDP per capita and "Life" is life expectancy. So then wouldn't it be correct to say component 2 represents "Agency" (or something similar)? In other words, societies which put more emphasis on a strong economy and medical advances, rather than social welfare and equality structures, tend to lead to happier people (or at least has a 10% weighting on variance in the data)?
Please correct me if I'm wrong, trying to get my head around this!
poor explanation
@thirumalainambi6068Excellent explanation !
@omaridbrayme1410So PCA can remove confounders by showing which variable accounts for most of the variation. That's perfect. I'd just go through and cut the weakest variables out of the model entirely in large models when my goal is explanatory power rather than predictive power.
@robertwilsoniii2048Thanks.
@ramonjaramillo9736but why u1 and u2 has same factors considered? and then how their eigen vactors and values are calcualted?
ideally u2 (second component of PCA) should have factors other than (GDP, SOCIAL and LIFE right??)
Perfect explanation of PCA. Thank you!
@LuckyStudy-z1vTHANK YOU, THAT WAS AN AMAZING VIDEO EXPLAINING THIS ALGORITHM
@hafsaotchiha4806add t sne as well please
@tanzeelurrahman1114turned it from a confusing matrix soup into something my simple brain can understand. Thank you !
@caioartusBest explanation for PCA Thanks a lot I searched a lot !
@nadamaher4104This is the most informative, concise material explaining the principles of the PCA method. Now I grasp how it works. Thanks!
@bernardokonski5122This is literary the best PCA explanation I've ever seen, tysm for making this video you are amazing!!!
@StrawberryJamBunnySuper useful. Thanks!
@nicolasgonzalezlExcellent video - in 6 mins I understood PCA better than hour long lectures. Thanks!
@sethagastyaBrilliant exposition, keep up the good work.
@ramdasmenon0908Sir a khouia lah i sehel 3lik π²π¦β€
@magnasethjourey1311Principal components are not correlated.
@lxbt638thank you finally understood the concept
@sachinsachin2293Thank you for the great explanation!
@suryatejaswi4751Indians be like: why did you drop my country saar! πππ.
@lyricass7810one question: could you, theoretically, for the second principal component(balance), have chosen other items of the original table chart? or do those have to be the same as in the first(power)? wouldnβt it have provided more information if we had calculated an eigenvector from other items, with other eigenvalues? or am i totally on the wrong track hereβ¦?
@EvilJared88Great video
@Taha-uc1ifExcellent work buddy. I really appreciate your efforts. Thank you so much for the video...π
@swapnilchavan7076the only guy out there who went directly to the point without any extra talking
@yaserhmoud8726nice to see morocco here
@ryanharington76243:03 sips tea: wow I'm learning nothing
@PhyAI33I have been searching for hours for the derivation of the PCA and all i got was the algorithm process only till i stumbled into this video! thank you so much!
@ahmedwesam7286exceiient
@uniquedream5413Rodriguez Paul Allen Helen Hernandez Matthew
@MathivpasIsakssonbeautiful! thx!
@Psychog-gGarcia Charles Moore Brenda Martin Sandra
@BurneJonesClaire-b1vJackson Jeffrey Williams Jeffrey Hall John
@LyttonDominic-s5lMartinez Steven Robinson Kevin Perez George
@FaradayDave-x2sTaylor Sharon Walker Nancy Moore Gary
@NancyMendozaiThank you
@varunkumar7237