93,593 reputation
8162327
bio website quantdec.com
location Northeastern US
age 14
visits member for 4 years, 4 months
seen 14 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 I can be reached through (outdated but still valid) links posted on my web site.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.


22m
reviewed Close Define feature for text classification using NLTK
22m
reviewed Close PCA on Binary Data
22m
reviewed Close Can PCA work when the number of observations is smaller than the number of dimensions?
23m
reviewed Close How do you conduct regression analysis using SPSS when there is more than one dependent variable?
23m
comment How to explain change of sign on regression coefficient when another variable is added to OLS model?
Those correlation coefficients are not "high" by any means! Did you notice the low VIFs?
26m
comment PCA correlation req normalizing data?
Your edit is helpful in helping us understand the problem, but as @gung remarks, the procedure it describes contradicts your assertion that you get different results with the two methods. We have to conclude you're doing something other than you describe--but we don't have enough information to know or figure out what the mistake is. Could you include specific information about what procedure you are actually carrying out?
32m
comment Reporting degrees of freedom for Welch t-test
I have not studied actual practice--which is why this is a comment and not an answer--but I would expect it to be based on judgment related to reporting significant figures. For relatively high df, often a change in the first decimal place would not change the p-value at all (to the level of precision reported), so rounding to an integer is fine. For very low df $\nu$ and extreme values of $t$, the derivative $|\frac{\partial}{\partial\nu}F_\nu(t)|$ can exceed $0.01$, suggesting in such cases that $\nu$ should be reported to only one less significant figure than $p$ itself.
1h
comment PCA correlation req normalizing data?
Search our site for "correlation," then: you should quickly discover that standardizing the variables does not change their correlation at all. Your question is confusing because it does not make it clear what you mean by "use both zscore and correlation matrix." Could you be more specific about what that procedure is?
2h
comment ratio of standard errors
I have never encountered any application where a "high standard error ... is desired." What exactly do you mean by "capture maximum variability" and why is that a good thing? And how does one go about achieving a large standard error--what are you proposing to vary to accomplish this?
2h
revised PCA correlation req normalizing data?
edited tags
2h
comment PCA correlation req normalizing data?
More useful discussion of this is available by searching our site.
2h
comment Monte carlo simulation in R
@Aksakal The Moshe reference is a good one, thank you. His (brief) chapter 4 discusses a simple, practical way to simulate queues like this one.
4h
comment Algorithm for uniform sampling with bounded replacement
Thanks. I'm still trying to get my head around what you want. It would help to fix the inconsistency in the last two bullets, which claim to apply to the same triple $(n,m,r)=(3,2,2)$.
4h
comment Algorithm for uniform sampling with bounded replacement
This sounds like a special case of the recently-asked question at stats.stackexchange.com/questions/129644.
4h
comment Monte carlo simulation in R
@mugen I mentioned the Matloff book in the text. It might be appropriate for those new to R who want another (but fairly similar) perspective on queue simulations. While writing this little simulator I found myself thinking a lot about how much I learned by studying the code in (the first edition of) Andrew Tanenbaum's text Operating Systems / Design and Implementation. I also learned about practical data structures, such as heaps, from Jon Bentley's articles in CACM and his Programming Pearls series of books. Tanenbaum and Bentley are great authors everyone should read.
4h
revised Monte carlo simulation in R
added 1134 characters in body
5h
comment Polynomial fitting for Shading Correction
Your first equation specifies, in the clearest possible terms, how to "build this function using the coefficients."
5h
comment What is the probability of observing some function given a gaussian process?
The t-test does not give the probability of observing a particular t-statistic: that would be zero. Likewise, for any Gaussian process the probability of observing any particular instance is always zero. What, then, are you trying to ask? How to write down the likelihood of some observations? How to conduct some kind of statistical test? If so, what are you trying to test?
5h
comment What means “CDF of one group does not cross the CDF of the other” from the dunn.test description?
I checked again. Glen_b's answer points out that "The step-functions can coincide, but the x-sample ecdf will never be above/left of the y-sample ecdf." That is a clear, straightforward (and graphically illustrated) response that directly answers your question.
5h
answered Monte carlo simulation in R