Timeline for Statistical computation in matrix. Rows before columns? riddle..
Current License: CC BY-SA 3.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 15, 2011 at 21:39 | comment | added | Barry Cipra | I ran the same test as Douglas (before seeing his answer) for (m,n)=(2,3), and got the same answers, 0.835 and 0.829. I'm not sure how Tangent is getting 0.958 and 0.968 here, unless it has something to do with what James Martin pointed out in the comments to the original problem. | |
| Aug 15, 2011 at 17:27 | history | edited | Douglas Zare | CC BY-SA 3.0 |
Tried to avoid saying first method versus second method.
|
| Aug 15, 2011 at 16:34 | comment | added | Douglas Zare | Perhaps what I called first and second disagree with your conventions, but I do not expect to see such a difference between my simulation's $0.991$ and your $0.926$. | |
| Aug 15, 2011 at 16:14 | comment | added | Tangent | About the numerical test. I'm ashamed to show it in a mathematical forum, but I ran a simulation in MATLAB, averaged over 10000 simulations. For $m=2, n=100$: Average for the 1st method (choosing column first) is $0.926$ and for 2nd method is $0.967$. I counted how many times each method was better: 1st>2nd: $0.18$, 2nd>1st: $0.59$, 1st==2n: $0.22$ For $(m,n)=(2,3)$: Avg for 1st: $0.958$ for 2nd: $0.968$. times 1st>2nd: $0.194$, 2nd>1st: $0.234$, 1st==2nd: $0.57$ For $(m,n)=(12,30)$: Avg for 1st: $0.88$ for 2nd: $0.9$. times 1st>2nd: $0.382$, 2nd>1st: $0.547$, 1st==2nd: $0.07$ | |
| Aug 15, 2011 at 12:11 | history | answered | Douglas Zare | CC BY-SA 3.0 |