Lecture 9

• Lecture 9
  • Video
  • PCA Maths
  • PCA: Steps

Video

link

• We use PCA to avoid Curse of Dimensionality
• entropy = measure of information, depends upon variability
• more the variance, btr, else no info
• say ht, wt data tha
• I rotate coordinate axes, how to assign some semantic now
• new x,y = x’, y’ is a linear combination of x, y
• we might not be able to attach some semantics to x’, y’

PCA Maths

• pca is interested only in direction, and not actual origin’ value, so we shift and make vector pass through 0

• covariance matrix is symmetrix

• column vector (3,2) is an eigen vector and 4 is an eigen value

PCA: Steps

• eigen vector give me the best direction

• mean adjust data
• get covariance matrix
• find eigen values and vectors
• arrange values from highest to lowest
• corresp vectors gice principal components