Publications by Wingfeet
Continuing Sync
I am continuing in Sync: How Order Emerges from Chaos in the Universe, Nature, and Daily Lifeby Steven Strogatz. To get a feeling on it, I was building a group of things which have only a minute influence on each other are able to synchronize their behavior. I got some more explanations how things fit together and adapted my calculat...
2247 sym 2 img
Ordinal data, models with observers
I recently made three posts regarding analysis of ordinal data. A post looking at all methods I could find in R, a post with an additional method and a post using JAGS. Common in all three was using the cheese data, a data set where only product data was available, observers were not there. The real world is different, there are o...
5910 sym
Simulation shows gain of clmm over ANOVA is small
After last post’s setting up for a simulation, it is now time to look how the models compare. To my disappointment with my simple simulations of assessors behavior the gain is minimal. Unfortunately, the simulation took much more time than I expected, so I will not expand it.BackgroundI have been looking at the ordered logistic mo...
4944 sym 4 img
Reshaping data
Preparing and reshaping data is the ever continuing task of a data analyst. Luckily we have many tools for it. The default tool in R would be reshape(), although this is so user friendly that a reshape package has been added too. I try to use reshape() (the function) because I feel it is a good tool, though with a somewhat cryptical m...
11372 sym
Bubble sort tuning
I was reading Paul Hiemsta’s blogpost on Much more efficient bubble sort in R using the Rcpp and inline packages, went back to his first post Bubble sort implemented in pure R and thought, surely we can do it better in pure R. So I cleaned inner loops, removed a recursion made some variations and finally made a big improvement ...
6859 sym
More bubble sort tuning
After last week’s post bubble sort tuning I got an email from Berend Hasselman noting that my ‘best’ function did not protect against cases n<=2 and a speed improvement was possible. That made me realize that I should have been profiling the functions rather than just look for inspiration in the code. So, that's what I wanted ...
4958 sym
Cars in Netherlands
I am looking for a new car. So when I saw there was an update on vehicles in Statistics Netherlands I just had to go and look at the data. So, I learned the brown is getting more popular, often the number of cars from a certain construction year is larger at six years of age than five years of age and lighter cars get more popular, es...
6121 sym 16 img
Distribution of car weights
Two weeks ago I described car data, among which weight distribution of cars in Netherlands. At that time it was purely plots. In the mean time I decided I wanted to model trends. As a first step of that, I decided to fit distributions for these data. I soon found out that most of the tools seem to be organized around individual objects as units....
6581 sym 4 img
Opel Corsa Diesel Usage
I wanted to extend my car weight distribution calculation of June 16 from only 2000 to years 2000 to 2013. Unfortunately, come Sunday afternoon the code seemed too slow and not even the beginning of a post. So, I went on to another calculation I was working on; scraping Diesel usage data from the web and examining how it varied. The ...
6951 sym 8 img
Faster calculation
Last week I decided to speed up my distribution fitting functions of two weeks ago. These were bold words. My try of Rcpp was a failure. Just plain optimization helped a bit better. Using the compiler package added a bit. (the compiler package does not compile, but a 10% speed increase at minimum costs is never bad).Rcpp (with inline)...
6807 sym