From: BHARATH on 17 Feb 2010 05:49 hi I need a cosine angle between vectors in the range [0,1]. here my vector is of 20,000 tuples. sholud i need to normalize it????? i know that theta=acos(dot(a,b))/norm(a)*norm(b) please tell me .. how to find it.... my aim is to measure the correlation between them...
From: Wayne King on 17 Feb 2010 06:44 "BHARATH " <bhushabhusha(a)gmail.com> wrote in message <hlghj2$6ab$1(a)fred.mathworks.com>... > hi > I need a cosine angle between vectors in the range [0,1]. here my vector is of 20,000 tuples. sholud i need to normalize it????? > > i know that theta=acos(dot(a,b))/norm(a)*norm(b) > > please tell me .. how to find it.... > my aim is to measure the correlation between them... Hi, you need to take the inverse cosine of dot(a,b)/norm(a)*norm(b). The way you have written it, you are taking the inverse cosine of the inner product of the vectors a and b and dividing that by the product of their norms. That is not the definition of theta. Also, see >>doc corrcoef Hope that helps, Wayne
|
Pages: 1 Prev: how can i test the "Encoder Input" from Real-Time Windows Target? Next: Resizefcn - old size |