[R] Panel.abline would not show beyond a certain slope value

[Aziz, Muhammad Fayez]

Yup. Latest version 2.15.0 for windows solved the problem alright! Thank you Uwe.

Regards,
Fayez
________________________________________
From: Uwe Ligges [ligges at statistik.tu-dortmund.de]
Sent: Monday, April 09, 2012 9:56 AM
To: Aziz, Muhammad Fayez
Cc: r-help at r-project.org
Subject: Re: [R] Panel.abline would not show beyond a certain slope value

On 09.04.2012 06:31, Aziz, Muhammad Fayez wrote:
>
> Hi,
>
> Please find the code and data following. Problem appears in lines 37 - 42 of the code. I am using R.2.13.0 on WinXP.

Works for me with a recent version of R (i.e. 2.15.0) / grid and lattice
(i.e. 0.20-6).
Please always try with recent versions of R and packages before asking.

Best,
Uwe Ligges



> Regards,
> Fayez
>
>
> PowerLawGraphsAblineQn.R‎
>
> library('reshape')
> library(lattice)
> library(igraph) # to use power.law.fit function
> library(latticeExtra) # to use panel.lmlineq in loglog xyplot
>
> File2Open = "C:\\Documents and Settings\\All Users\\Documents\\Desktop\\Fayez\\RPractice\\PowerLawGraphsAblineQnData.txt"
> DataTable = read.table(File2Open, header = TRUE, sep = "\t")
> md<- melt(DataTable, id.vars = c('Range', 'TheValue', 'TheNo')) # removed 'DegType' column 030312
> md$variable<- as.numeric(substr(md$variable, 9, nchar(as.character(md$variable))))
>
>
> mypanel4loglog<-
>      function(x, # x is the variable column in melted data, equals the Domain Nos
>          y, # y is the value column in melted data, the degrees
>          ... # Rest of the arguments
>      )
> {
>
>      kfreq<- table(y); # compute frquency hash table of y, the values
>      k<- 1:max(y)
>      for (i in k) {
>          ichar<- as.character(i) # convert to match the names(freq), the character-based hash key of freq, which is the value
>          if (!(ichar %in% names(kfreq)))
>              kfreq[ichar]<- 0
>      }
>      sortedkeys<- as.character(k)
>      kfreq<-  kfreq[sortedkeys]
>      pk<- kfreq / length(y)
>      logk<- log(k)
>      logpk<- log(pk)
>
>      logpk[logpk == -Inf] = "" # remove the -Inf or log(p(k)) = 0 values for lm function, NULL is 0-length so use "" instead that has length of one null character
>      logpk<- as.numeric(logpk) # "" converts all values to character, lm needs numeric
>      print(rbind(logk, logpk))#        write.table(rbind(k, kfreq, pk, logk, logpk), paste(FilePath, "\\data", sep=""), sep = "\t", append = TRUE)
>
>      panel.xyplot(col="blue", logk, logpk, type = c('p', 'r'))
>
>      lm.r = lm(logpk ~ logk)
>      panel.abline(coef=c(-4.847634, -1.037480)) # ->->->  This gets drawn
>      panel.abline(coef=c(-4.847634, -1.037481)) # ->->->  This doesn't get drawn
>      print(coef(lm.r)) # -4.847634   -1.349699 ;  -3.377894   -1.498693
>
> } # end mypanel4loglog
>
> myprepanel4loglog<-
>      function(x, # x is the variable column in melted data, equals the Network's ages
>          y, # y is the value column in melted data, the degrees
>          ... # Rest of the arguments
>      )
> {
>          FreqTable<- as.data.frame(table(y))
>          FreqsVector<- sort(FreqTable$Freq)
>          Min<- FreqsVector[1] # first element - the lowest value frequency
> #            print(c(Max2, length(y)))
>          list(ylim = c(log(Min / length(y)), 0), xlim = c(0, log(max(y)))) # log(p(k)) is always -ve as p(k) is decimal, so max(log(p(k)) is 0
> } # end myprepanel4loglog
>
> print(xyplot(value ~ variable | Range,
>      data = md,
>      xlab = "log(k); Panel = Range",
>      ylab = "log(p(k))",
>      main = "log(k) vs. log(p(k))",
>      groups = Range,
>      pch = 20, # dots instead of circles
>      panel = mypanel4loglog,
>      prepanel = myprepanel4loglog, # to set the scale of k and pk
>      scales = list(x = (relation = "free"), y = (relation = "free")), # necessary to make x-axis in each panel adjustable according to k
> ))
>
>