meta user
network profile
|
May 20 |
comment |
Product of two independent random variables (...) are disjoint and will be respected once we transform back via $\exp$. The mixture weight is obtained from the proportion of negative samples. I fear I must have misunderstood your comment or am overlooking something. I admit this comment and the previous were almost completely off-the-cuff. Hopefully they have not introduced too much noise. |
|
May 20 |
comment |
Product of two independent random variables @whuber: Apologies in advance for only having given your comment a cursory reading. I was a little unsure of what you're driving at. To (slightly) clarify my suggestion: If $Z = \xi \psi$, decompose as $Z := Z_{+} - Z_{-}$ where $Z_{+} = \xi \psi_{+}$ and $Z_{-} = \xi \psi_{-}$. Note that $\psi_{+} \geq 0$ iff $Z_{+} \geq 0$, so in particular, the number of negative samples is a sufficient statistic for $\mathbb P(Z < 0)$. Now apply deconvolution to $Z_{+}$ and $Z_{-}$ separately. I don't see how one estimate can contaminate the other since the original supports (...) |
|
May 20 |
reviewed | Close Possible to create random slope model with fixed intercept in SPSS? |
|
May 20 |
reviewed | Close SPSS: Plot a multinomial logistic regression |
|
May 20 |
reviewed | Leave Open Hilbert curves: bounds / probabilities of preserving neighbors |
|
May 20 |
reviewed | Leave Open Product of two independent random variables |
|
May 20 |
comment |
Product of two independent random variables @Xi'an: Nice links. I sure hope that $\psi \geq 0$ almost surely...though we can recover from a seemingly fatal violation of this condition by decomposing as $\psi = \psi_{+} - \psi_{-}$ and considering the pieces separately. |
|
May 6 |
comment |
Foundamental limit theorem of Markov chains with higher order chains? Crossposted. (Please don't do this in the future. Choose the single best site for your question.) |
|
Apr 30 |
reviewed | Approve suggested edit on how to estimate CTR (ctr-click-through-rate)? |
|
Apr 30 |
comment |
Asymptotic property of tuning parameter in penalized regression It may also help for you to specify which asymptotic properties you are interested in. For example, (as you are likely to know) there is a significant difference between asymptotic consistency of the parameters and asymptotic consistency of model selection. There are even some results to the effect of not being able to get both of them together! |
|
Apr 30 |
comment |
Asymptotic property of tuning parameter in penalized regression Hi, Nick. Are you familiar with the paper W. Fu and K. Knight (2000), Asymptotics for lasso-type estimators, Ann. Stat., vol 28, no. 5, 1356-1378? This paper doesn't address your question head-on, but I mention it because it gives a simple framework to see what actually matters, while also allowing for considerable generality. They use a fixed sequence $\lambda_n$, but you can adapt their Theorem 1 to a sequence dependent on the data (e.g., via cross-validation) without too much difficulty. |
|
Apr 26 |
awarded | Guru |
|
Apr 26 |
reviewed | Approve suggested edit on Pre to Post Repeated measures for categorical dependent variable and more than one independent variables both continues and catgorical |
|
Apr 26 |
reviewed | Leave Open What is the value of assortativity for k-regular graphs? |
|
Apr 26 |
reviewed | Leave Open How to compare three different replacements for eggs in baking? |
|
Apr 26 |
reviewed | Close How to constrain parameters in reression models in STATA? |
|
Apr 26 |
reviewed | Close Balanced chromosomal translocation |
|
Apr 26 |
reviewed | Leave Open Will data mining be useful for me? |
|
Apr 26 |
reviewed | Leave Open Correlation Clustering |
|
Apr 26 |
reviewed | Close How to create an ADBUDG Economic Marketing Model with R |
