1,325 reputation
3931
bio website
location
age
visits member for 1 year, 7 months
seen Aug 12 at 20:10

Jul
15
comment “I have found a dead body on my car.”
Related: math.stackexchange.com/q/513317/56801
Jul
15
comment Poker blind interest equation
$(4300/50)^{(1/9)}-1=0.6403\dots$
Jul
2
awarded  Curious
May
14
accepted Is a log-normal distribution uniquely determined by its moments or not?
May
14
asked Is a log-normal distribution uniquely determined by its moments or not?
May
14
comment How does one 'correct' a table that doesn't add up to $100\%$?
I could do that, but I must admit that I'm not very familiar with centroids, hypercubes and hyperplanes. That way it might become obvious that I don't know much at all. :) I thought that the question as such must have been thought of by many people. For example those making graphs and pie charts and such.
May
13
revised How does one 'correct' a table that doesn't add up to $100\%$?
added 279 characters in body
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
@EmmadKareem Because if the error would be $-1\%$, some $\text{E}(y_i)$s (e.g. those belonging to $x_i=0$) could become less than zero. That indicates that that procedure cannot be right.
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
Thank you. The first paragraph, I think, captures my problem. However, the problem is that the symmetry (as mentioned in your second paragraph) applies only to numbers from $1$ through $99$. $0$ and $100$ are different. And in the tables that I have to work with, there are $0$s (as mentioned in the question). :(
May
13
revised How does one 'correct' a table that doesn't add up to $100\%$?
added 260 characters in body
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
For example take $10\%$, $80\%$ and $11\%$. Total $101\%$. Just scaling those down would obviously yield $100\%$. But $80\%\cdot 100/101=79.2079\ldots$ will now round to $79\%$ instead of $80\%$. So, this cannot be right.
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
Yes, but I think that in some cases that $\text{E}(y_i)$ won't round to $x_i$ anymore. That's why I came up with the explicit prior distribution.
May
13
revised How does one 'correct' a table that doesn't add up to $100\%$?
added 67 characters in body
May
13
revised How does one 'correct' a table that doesn't add up to $100\%$?
added 58 characters in body
May
13
revised How does one 'correct' a table that doesn't add up to $100\%$?
edited tags
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
@RossMillikan I'm fully aware that any $y$s don't change the $x$s. I'm only interested in reasonable estimates of the $y$s. The $x$s stay exactly what they were.
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
I that case, I would be looking for an answer that gives $100/3\%$ for each. (I'm not interested in 'higher'.)
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
I don't have the unrounded $y$s available. I want to estimate them. And this is not for election purposes. I only want a reasonable (actually: the most reasonable) unrounded table that rounds to the original one and adds up to $100\%$. Is my question that unclear, I worry.
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
@RossMillikan I know I cannot recover the $y$s. But given my assumptions, they could be estimated.
May
13
comment How does one 'correct' a table that doesn't add up to $100\%$?
I only want to reconstruct the ('expected') $y$s. Obviously, every $y$ needs to round to its $x$. Also, obviously, because the $y$s are exact, they add up to $100\%$. Why am I failing to bring this point across?