Lecture 11

• Lecture 11
  • Video
  • SVD
  • Inv Scree Test
  • Gram Schmidt Orthonormalization

Video

link

SVD

• we decomposed a matrix to 3 orthogonal matrix aa’ = a’a = i

• purpose of svd is not to find a but to dimesnionality reduce

Inv Scree Test

• graph bana
• y pe percentage variance
• x pe pc1, pc2, pc3…. pcn for n dimensional data which are ordered
• pc1 = best variability, pcn = lease

• slope of curve is pehle mast, then same hora aage badh ke
• it helps to decide ki kitni dimens reduce karu

Gram Schmidt Orthonormalization

• dimension of S matrix tells me ki kitni dimensions I am taking
• S has to be reduced diagonal matrix