91,857 reputation
8158322
bio website quantdec.com
location Northeastern US
age 14
visits member for 4 years, 3 months
seen 12 mins ago

Consultant (environmental and spatial stats a specialty), expert witness, and teacher. I can be reached through (outdated but still valid) links posted on my web site.

Twitter: @WilliamAHuber // ASA-P website: http://amstatphilly.org/


Why waste time learning, when ignorance is instantaneous?

--T(iger) Hobbes.

For any complex problem there is a simple solution. And it's always wrong.

--[Mis?]attributed to H.L. Mencken by Dava Sobel, Longitude.


1h
comment How do residuals relate to the underlying disturbances?
This is the right approach. What it needs in addition is an argument that the diagonal elements of $H$ are typically "small". That is made by showing that the trace equals the number of independent variables (including the intercept, if any)--which is immediate from the fact it's a projection matrix. Note that this result is independent of any distribution assumptions on the individual $\varepsilon_i$: they need not be Normal. It is also independent of any actual formula for $H$; it is a consequence of a count of dimensions.
1h
comment Matlab: How to implement inverse of determinant of covariance-variance matrix
There are no taus anywhere in that slide. It merely gives the likelihood for a multivariate Normal distribution. Perhaps you mean "Gamma" ($\Gamma$), but there is no determinant of the inverse of $\Gamma$ in that slide, either. Did you notice how on the next slide optimizing this likelihood is referred to as a "difficult optimization problem"? A simple automatic black-box search method is not going to be a realistic solution.
1h
comment What is the value of “X” in a regression equation when dealing with a time series?
Confidence intervals for such a fit would be meaningless: those data certainly don't follow such a trend and it's not even a good qualitative description.
1h
comment Matlab: Unable to understand expectation operator in the context of a vector
I'm sorry, but I haven't a clue what you are asking.
1h
comment Is it unusual for MEAN to outperform ARIMA?
FWIW, his LinkedIn page--which he maintains--lists both his PhDs in statistics. But the argument is pointless: claiming that somebody is not a statistician because their degree might not specifically be in statistics has little value and is beside the point here. (Until very recently most people whose career was in statistics had degrees in other fields because there were few statistics programs available.)
3h
comment Is it unusual for MEAN to outperform ARIMA?
I really like the perspective you provide on forecasting and the evidence you adduce to support it, but elements of this answer are objectionable because they read too much like a rant against "statisticians" or perhaps against formal statistical training--and are flat out wrong. For instance, Makridakis' advanced degrees are in (guess what?) statistics, that's what he teaches, and that's what he does.
4h
comment Matlab: Unable to understand expectation operator in the context of a vector
Although I have to guess what your notation means, it may be worth remarking that in a model, the $y(t)$ would ordinarily refer to random variables, not arrays of numbers, and the expectation is in the sense of random variables, not averages of arrays of numbers.
5h
reviewed No Action Needed Multiple comparisons of proportions using complex survey data using SAS
6h
reviewed Reviewed Two-Step estimation
6h
comment Two-Step estimation
So that this question can be answered, please describe your data and the kinds of "purchase equations" (whatever they might be) that you have in mind.
6h
comment How would you explain covariance to someone who understands only the mean?
@Tjorriemorrie This isn't about statistics, it's about geometry: You can construct plenty of rectangles whose sides are not parallel to the coordinate axes.
6h
comment Formula the conditional probability of marbles
It that's your question, then you need to edit the post. Questions here need to be understandable without having to refer to another site or reference.
6h
comment Formula the conditional probability of marbles
That paper gives a specific probability distribution for $(g,h)$ and it also explicitly answers your question.
6h
comment Formula the conditional probability of marbles
It seems to me, then, that the count of your "repeated connections" is just $g-h$. That doesn't seem to leave anything else to answer until you can specify the probability distributions for your choices of $g$ and $h$.
6h
revised Formula the conditional probability of marbles
edited tags
6h
comment Formula the conditional probability of marbles
We need more details to answer this question. How exactly are the marbles chosen "randomly" in order to assure $g\gt h$? Are the connections made randomly? If so, precisely how? How are repeated connections counted? For instance, in the illustration are there two repeated connections (for blue marbles 4 and 6), three (for the red lines), or even five (for green marbles 3, 5, 7, 9, and 2)?
7h
comment Salary Range Determination When Given Percentiles
@Gung certainly the parent distribution can be Normal: it will work perfectly fine provided the chance of an unrealistic salary is negligible. However, it is rare that any set of salary data looks like a sample of a Normal distribution.
7h
comment Choosing samples of different sizes - considerations?
@John's comment is well to the point. It is hopeless to expect better results by eliminating information! It appears your problem is really about clustering and not about sampling at all. But to make progress you would need to provide details of what you are attempting to do and what the results have been so far.
7h
comment Choosing samples of different sizes - considerations?
"Each sample has an N of maximum 1000" and "1000 samples from each population" both suggest many more than 1000 units can be sampled in total. Regardless, when introducing the result of a calculation in an answer it will help your readers to explain how you arrived at it.
8h
comment Choosing samples of different sizes - considerations?
Where does the 7.3% figure come from?