rosser

801 reputation

412

bio	website
	location	Dublin, Ireland
	age
visits	member for	1 year, 7 months
	seen	Mar 5 at 2:13
stats	profile views	88

Psychiatrist, Trickcyclist, Marathoner

	801 reputation	bio	website		visits	member for	1 year, 7 months
412 badges		location	Dublin, Ireland		seen	Mar 5 at 2:13

summary answers questions tags badges favorites bounties reputation activity

33 Comments

suggestions reviews revisions comments badges posts accepts all

Jul 12	comment	How to deal with ceiling effect due to measurement tool? Tobit? or not tobit?
Jun 29	comment	GWAS and Statistical theory - does the likelihood of a detectable main effect decrease with complexity? corrected - but kind of not the point
Jun 28	comment	When/where to use functional data analysis? I think it depends on the density of your data, whether you consider it closer to repeated measures or a time series - is that wrong?
Jun 21	comment	What are the assumptions of ordinal mixed effects logistic regression? The reviewer is not a ststistician and will definitely not get that.
Jun 19	comment	What are the assumptions of ordinal mixed effects logistic regression? I know that and u know that, my question is how to make him/her understand that. Do I just go around handing out copies of gelman & hill??
Mar 27	comment	How to choose the best of two highly correlated predictors in cox proportional hazards regression Sorry, yes .. and I did.. thanks
Mar 5	comment	If you use 10-fold cross validation, which tree is representative? Thanks. I am a moron.
Mar 5	comment	Binary Classification of Multiple Groups Why not just use a logistic regression with Dead/Alive as the dependent and have a dummy variable for Drug A,B,C with your patient characteristics as covariates, which will give you an odds ratio?
Mar 5	comment	One sentence explanation of the AIC for non-technical types @Dilip - so far past the line, you can't even see the line anymore -- lol, thanks. I think I'll use something between Peter's and my own.
Jan 9	comment	How do I report error from imbalanced data in a random forest algorithm? However, on a similarly unbalanced hold-out sample (this time 5:1 non-cases to cases) I am having the same difficulty as you would expect - the machine is very specific, but not very sensitive.
Jan 9	comment	How do I report error from imbalanced data in a random forest algorithm? I used WEKA's facility for random resampling in the preprocessing stage which samples with replacement from both groups giving you a (fairly balanced) dataset. So from 151 cases it went from 100:51 to 78:73. I then developed a random Forest on that set and the results above are just 10-fold cross validation on that data set. I didn't interfere with the CV process after resampling the data. The 20 predictors were selected based on regression modelling and the VIMP of a random forest as well as domain knowledge. I didn't use a hold-out sample for the results above, although I have 99 case.
Jan 9	comment	How do I derive principal components taking account of repeated measures? Sorry - away due to illness - yes you are correct in your assumptions. Thanks for any help.
Dec 4	comment	Why the infrequent use of machine learning techniques in translational biomedicine? Yeah, I think I got that. However, from my point of view statistical methods are neither pure nor dirty, just the application of logic to data. If you want a pill to cure something, then you need to understand the interrelationships and take that to the molecular biology lab. However, If you just want to make a prediction using black box (NN/RF) or decision (CART) methods, what's the problem? You might even gain insight. Is it any deeper than snobbery?
Dec 4	comment	Building background for machine learning for CS student Also, for an exceedingly gentle "ramp" see the current video lectures on machine learning by Andrew Ng at Stanford. They provide a fairly sound introduction to Hastie or Bishop. ml-class.org/course/class/index
Dec 4	comment	Neural network model to predict treatment outcome Reposted the followup question as I had it written here with minor tweaks.
Nov 25	comment	How to make a randomForest algorithm cost-sensitive? biomedcentral.com/content/pdf/1471-2105-10-S1-S22.pdf Using random forest for reliable classification and cost-sensitive learning for medical diagnosis
Nov 25	comment	How to make a randomForest algorithm cost-sensitive? Looks like the bst package in R will accept weights for false positive and false negatives. I'm going to leave the question open though, It might be useful to others.
Nov 10	comment	How do I analyse data with a ceiling effect? censReg package with plm.data looks promising
Nov 10	comment	How do I analyse data with a ceiling effect? I wonder is there a way to combine these ... there doesn't seem to be an established way to do a repeated measures Tobit model in the package VGAM. That would seem to be the most elegant solution. Previously, the (awful) literature has used raw change scores across 2 timepoints, with presumably significant insensitivity due to regression to the mean.
Nov 10	comment	How do I analyse data with a ceiling effect? However, at baseline there's quite a spread below 30, so it is detecting "disability", however, when people return to "normal" functioning after treatment, they all arrive at 28-30 or thereabouts.