Reputation
Next tag badge:
1197/1000 score
195/200 answers
Badges
9 191 381
Newest
 covariance
Impact
~2.9m people reached

20h
comment Approaches on how to analyze product allocation (for multiple warehouses) by region in R?
Questions 2, 3, and 4 cannot be answered without further information. Overall it is important to explain how these data were obtained and what they represent. Are they a sample of something? How were the individual records in the sample selected? How were the values in those records measured? What do those values mean? Please edit this post to narrow its focus to one type of question and to include this essential information.
20h
comment Contrast 3 is orthogonal to Contrast 1, Contrast 2
This question is under-specified, since there are two dimensions of vectors orthogonal to both C1 and C2. What additional criterion must Contrast 3 satisfy?
20h
comment markov chain question
According to the information given, this is not a Markov chain and the (obvious) answer is $1/6+3/6$. The lack of any stated relationship between distances and probabilities and the absence of any indication that this is a process (instead of a single, one-time choice) lead me to suspect you have not reproduced the original question accurately.
20h
reviewed Leave Open Why are there two spellings of “heteroskedastic” or “heteroscedastic”?
20h
reviewed Close High asymmetric binary variables
20h
reviewed Leave Open How to simulate from a log-copula function?
20h
reviewed Leave Open KS, RMSE tests in R
20h
comment KS, RMSE tests in R
If those GEV parameters were estimated from the data, then this use of the KS test is invalid and will return p-values that can be much too small.
20h
reviewed No Action Needed Nonlinear dimensionality reduction (sample size is smaller than number of features)
20h
reviewed Leave Closed Opportunities in machine learning and computational intelligence
1d
comment Chances of a reducing probability
Chances of what, exactly? That "it felt extraordinary?"
1d
comment Opportunities in machine learning and computational intelligence
What is your question?
1d
comment What does raising frequencies to a power do when calculating probabilities?
Give us a hint at least! What's the context? What is this part of the code working on?
1d
comment Editing a Macro (principal components and cross validation)
I'm sorry; we're neither a code review nor a coding site.
1d
comment Recovering number of planets in a sample
The event to be worried about is simpler and perhaps far more likely than that: what is the chance that some star with low $p$ has a large number of planets? That possibility, if it is sufficiently large (and that will depend on the proportion of low-$p$ stars), can elevate the UCL a large amount unless you can bound the numbers of planets--but your question in its current form posits nothing about that. This suggests your problem is, at present, under-specified.
1d
reviewed No Action Needed Interpreting random effect variance in glmer
1d
comment Recovering number of planets in a sample
Fair enough. It strikes me that if even a single one of the $p_i$ is extremely small, then the upper confidence limit would be extraordinarily large by virtue of the possibility that such a star might host many planets, none of which would likely be detected. It could therefore help to specify something about the likely numbers of planets around such stars. You should also watch out for the possibility that your observations are not independent.
1d
comment Multiplicative error and additive error for generalized linear model
So that others are not confused, the notation "$E(Y)=\exp(X\beta)$" is not a correct way to describe a distribution, nor is it a correct way to describe a Gamma variable: it merely re-states the link function. I suspect you meant something like $Y\sim\Gamma(k, \theta)$.
1d
comment How do you use mathematical notation to write out an equation for a GLM with a gamma distribution and a log link?
You can find plenty of examples of how to do this for GLMMs and various links by searching our site‌​. Some of the first hits I found include stats.stackexchange.com/questions/20835 (binomial with logit link) and stats.stackexchange.com/a/64039/919 (binomial with a custom link). Just plug in your log link and Gamma distribution to answer your own question. The example in the question at stats.stackexchange.com/questions/114046 is incorrect.
1d
comment Recovering number of planets in a sample
Thank you. The meaning of $p_i$ is still unclear, though. Is it the chance, given the star has a planet, of detecting at least one planet at star $i$ or an independent chance per planet at star $i$ of being detected? Is there some chance of falsely detecting planets at stars without them? How do you know the values of each $p_i$? (I suspect you do not know them to high accuracy--most likely they are estimated from properties of the star, its distance to us, and the method used to detect planets at that star--so your confidence interval ought to incorporate that component of uncertainty.)