Since we have freeze of Debian Squeeze, new features are not accepted any more and only important bug fixes can be excused and uploaded into testing, latest version 0.7.0.4-1 of gedit-r-plugin will not make it.
However the same packages that will be uploaded to unstable can be obtained from my private mirror
wget http://www.kaduk.net/~mateusz/gedit-r-plugin/kaduk.asc ; cat kaduk.asc | apt-key add -
echo "deb http://www.kaduk.net/~mateusz/gedit-r-plugin ./" >> /etc/apt/source.list
apt-get update
apt-get install gedit-r-plugin
Categories: Uncategorized.
I implemented cross-correlation function in order to correct for missing samples.
meanx <- function(x,wx,tau) {
return(sum(x[(tau+1):length(x)] * wx[(tau+1):length(wx)]) / sum(wx[(tau+1):length(wx)]));
}
meany <- function(y,wy,tau) {
return(sum(y[1:(length(y)-tau)] * wy[1:(length(wy)-tau)]) / sum(wy[1:(length(wy)-tau)]));
}
sccf <- function(x,y,tau) {
wx = as.numeric(x!=0);
wy = as.numeric(y!=0);
w = as.numeric(wx[(tau+1):length(wx)] & wy[1:(length(wy)-tau)]);
q = sum(w*(x[(tau+1):length(x)] - meanx(x,wx,tau)) * (y[1:(length(y)-tau)] - meany(y,wy,tau)));
d = sqrt(sum((w*(x[(tau+1):length(x)] - meanx(x,wx,tau)))^2)) * sqrt(sum((w*(y[1:(length(y)-tau)] - meany(y,wy,tau)))^2));
r = q/d;
return(r);
}
mccf <- function(x,y,tau) {
rlist = 0;
for(jtau in (-tau):tau) {
if(jtau < 0) {
tmp = y; y = x; x = tmp;
jtau = -jtau;
}
r = sccf(x,y,jtau);
rlist <- c(rlist,r);
}
return(cbind(rlist[2:length(rlist)], (-tau):tau));
}
Categories: R.
Recently I packaged RGedit for GNU/Linux – Debian and it has been uploaded into sid gedit-r-plugin.
I also joined RGedit project at this time as second developer. My first intention was to maintain essential patches on upstream level, but in a spare time I also plan to contribute in
- Code clean-up (with focus on porting glade->gtkbuilder)
- New docbook documentation
- New features development
Categories: GNU/Linux.
This is the implementation of ASCA for two factorial design(dosage and time)
# ASCA
anova_model <- function(input, Treatment, Subject, Time) {
# Replace NAs
input_w = is.na(input);
input[input_w] = 0;
# Mean center input
Xc = apply(input,2,function(col) col-wmean(col));
Xc[input_w] = 0;
# Average over k to get X.k.
X.k. = rep(0,max(Time)*ncol(Xc));
dim(X.k.) <- c(max(Time),ncol(Xc));
for(k in unique(Time)) {
# Take into account number of treatments per subject
colsum = colSums(Xc[Time == k,]);
if(sum(colsum) != 0) { # Unless not available sample for all subjects
n = sum(cbind(Treatment,Subject,(Xc[,1]!=0))[Time == k,][,3]); # Number of subjects used to calculated sum for given k
colsum = colsum / n;
}
X.k.[k,] = colsum;
}
# Average over hk levels to get Xhk.
Xhk. = rep(0,max(Time)*max(Treatment)*ncol(Xc));
dim(Xhk.) <- c(max(Time)*max(Treatment),ncol(Xc));
i = 0;
for(h in unique(Treatment)) {
for(k in unique(Time)) {
i=i+1;
colsum = colSums(Xc[Treatment == h & Time == k,]);
if(sum(colsum) != 0) { # Unless not available samples for all subjects
n = sum(cbind(Treatment,Subject,(Xc[,1]!=0))[Treatment == h & Time == k,3]); # Number of subjects per hk level, max=5
# Do the weighting
weight = sqrt(n);
# Calculate average
colsum = (colsum / n) * weight;
}
# Put back intro matrix
Xhk.[i,] = colsum;
}
}
# Calculate interaction k&hk
substraction = X.k.;
for(i in 1:((dim(Xhk.)[1]/dim(X.k.)[1])-1)) {
substraction = rbind(substraction,X.k.);
}
Xinteraction = Xhk. - substraction;
# Calculate redisudals
Xhki = Xc;
substraction = Xhk.;
for(i in 1:((dim(Xhki)[1] / dim(Xhk.)[1])-1)) {
substraction = rbind(substraction,Xhk.);
}
Xresidual = Xhki - substraction;
return(list(X.k. = X.k., XinteractionHK=Xinteraction, XresidualHKI=Xresidual));
}
Categories: R, Statistics.
tc filter add dev eth0 parent 1:0 protocol ip prio 1 \
u32 \
match ip dst 192.168.1.100
match ip dport 0x400 0xB1DE flowid 1:1
To begin with first line is typical to all filtering rules, second stands for u32 filter usage and third for matching dst IP address. However the most interesting is 4th line which matches destination port range from 1024 to 20001.
The rule is the same as with IP address masks. The first 0×400 hex value stands for 1024 and the second is the mask. The mask is calculated (dec 20001) 0x4E21 XOR 0xffff = 0xB1DE.
Categories: GNU/Linux.
To begin with download GeditLatexPlugin or if on debian
apt-get install gedit-latex-plugin
Secondly run gedit enabled LaTeX plugin and configure it, go to build profiles and change
R Sweave %e.Rnw
to
R CMD Sweave %f
After this changes you use R Sweave button to compile and to include your R code in LaTeX use
\begin{Scode}{fig=TRUE,echo=FALSE}
... R code ...
\end{Scode}
Categories: GNU/Linux.
