Publications by Wingfeet
For descriptive statistics, values below LLOQ set to …
That is what I read the other day. For calculation of descriptive statistics, values below the LLOQ (lower limit of quantification) were set to…. Then I wondered, wasn’t there a trick in JAGS to incorporate the presence of missing data while estimating parameters of a distribution. How would that compare with standard methods such as imputi...
7976 sym 14 img
Finding a jump in data
The other day I saw a plot with some data with a clear jump in it. So I wondered if it were possible to determine the position of that jump. Data with a jump is easy enough made:library(ggplot2)n mydata mydata$y .5,0,1)qplot(x=mydata$x,y=mydata$y)Now to find where that jump was. After some consideration, the most simple way seemed ap...
4110 sym 6 img
A look at strucchange and segmented
After last week’s post it was commented that strucchange and segmented would be more suitable for my purpose. I had a look at both. Strucchange can find a jump in a time series, which was what I was looking for. In contrast segmented is more suitable for occasions where rates of effect. This made me think of a post on climate change...
5216 sym 6 img
Earthquakes in Netherlands
In the Netherlands we have Natural Gas. Unfortunately winning this gas seems to cause some quakes. As quakes go, they are not strong. However, our buildings are not made to resist quakes, before 1986 they were unheard of, so there is some damage. It is now predicted they could get stronger and more frequent. This caused a bit of a st...
6360 sym 4 img
Occurrence of Alzheimer deaths in Netherlands
I wanted to know a bit more about the number of people suffering from Alzheimer. It has been getting more attention and it runs in my family. From CBS (Statistics Netherlands) I pulled the data regarding the number of people dying from it. I am afraid that means some Dutch words but that’s been kept low. Obviously death is not the worst of thi...
4474 sym 8 img
More sequential testing for triangle tests
I looked before at triangle tests and at sequential testing in triangle tests (blog entry). In the latter post it was demonstrated that a sequential test is possible, without costs in desired error of the first kind. The latter because the desired alpha (e.g. 5%) is never exactly obtained in any test based on the binomial distribut...
6252 sym 4 img
Ordinal Data
I expect to be getting some ordinal data, from 5 or 9 point rating scales, pretty soon, so I am having a look ahead how to treat those. Often ANOVA is used, even though it is well known not to be ideal fro a statistical point of view, so that is the starting point. Next stops are polr (MASS), clm (ordinal) and MCMCoprobit (MCMCpack). ...
13141 sym 6 img
Ordinal data with JAGS
Last week is had a look at the standard R routines for estimating models for ordinal data. This week, I want to have a look at JAGS for examining the same data. To be honest, most of it is taking an example (inhaler) and removing code. To my surprise this was almost as difficult as adding complexity in models. I did add some output from JAGS. Th...
4524 sym
More ordinal data display
The past two weeks I made a post regarding analyzing ordinal data with R and JAGS. The calculations in the second part made me realize I could actually get top two box intervals out of R. This demonstrated here. For that I needed the inverse logistic transformation. This I pulled out the ARM package, and I now realize the latter al...
3992 sym 2 img
Sync
I am listening to the audiobook Sync: How Order Emerges from Chaos in the Universe, Nature, and Daily Lifeby Steven Strogatz which I got from Audible. Obviously a mathematical book is not ideal to listen to, but lacking illustrations I can make them myself. On top of that, you learn more from programming than listening.He mentioned ...
3661 sym 8 img