Re: MSE : correlation regression

Posted by Chinni18 on Jun 06, 2012; 1:16pm
URL: http://discussion-forum.276.s1.nabble.com/MSE-correlation-regression-tp7577509p7577513.html

For Q.3
The minimum value of f(x)=x^x is at 1/e
Let f(x) = y
y = x^x
=> log y = x* log x
=> (1/y) (dy/dx) = logx + 1
dy/dx = (x^x) (log x + 1 )
For minimum, put dy/dx = 0
=> log x + 1 = 0
=> x = e^ -1 = 1/e
If you substitute this value in the second derivative of y, you'll get > 0, therefore it is indeed a point of minimum
Option a