Problem Rounding Up

[EvmARodCur]

When trying to use the round suppress, it sometimes rounds with a very big negative objective value. To show this, using the r code "test_code"

`
###fill
test<-read.table(file="CSV_CODIF.csv" , sep="\t", header=TRUE)
###fill

names <-colnames(test)

tot_code<-c("C1_1","C2_1","C3_1","C4_1")
dirname <-"tauargus_files"
filename <-"ex1"
dir_temp<-paste("/",dirname,"/",filename,".csv",sep="")

###fill
file_hrc=c("U:/anonimizacion/r tau argus/comprobacion/tabla_a_codificar/hrc_C1.hrc")
###fill

list_hrc<-setNames(file_hrc,"C1")

library(rtauargus)

###fill
loc_tauargus<-"C:/Users/ext-arodcur/Desktop/TauArgus4.2.4b2/TauArgus4.2.4b2/TauArgus.exe"
###fill

options(rtauargus.tauargus_exe = loc_tauargus)

ex1<- tab_rtauargus(
test,
dir_name = dirname,
files_name = filename,
explanatory_vars = names[-length(names)],
safety_rules = "FREQ(5,20)",
value= "OBS_VALUE",
freq = "OBS_VALUE",
totcode = tot_code,
suppress= "RND(1,5,0,1000,0)",
hrc=list_hrc,
verbose = TRUE,
)

###fill
test<-read.table(file="CSV_CODIF_2.csv" , sep="\t", header=TRUE)
###fill

names <-colnames(test)

tot_code<-c("C1_1","C2_1","C3_1","C4_1")
dirname <-"tauargus_files"
filename <-"ex1"
dir_temp<-paste("/",dirname,"/",filename,".csv",sep="")

file_hrc=c("hrc_C1.hrc")
list_hrc<-setNames(file_hrc,"C1")

ex1<- tab_rtauargus(
test,
dir_name = dirname,
files_name = filename,
explanatory_vars = names[-length(names)],
safety_rules = "FREQ(5,20)",
value= "OBS_VALUE",
freq = "OBS_VALUE",
totcode = tot_code,
suppress= "RND(1,5,0,1000,0)",
hrc=list_hrc,
verbose = TRUE,
)`

if we run the data from the csv CSV_CODIF
CSV_CODIF.csv

that has a hierarchy in the first column in the file hrc_C1 (must be changed to hrc format):
hrc_C1.csv

we get that negative objective value:

with rounded values far from the original values:

Meanwhile if we do the same with CSV_CODIF_2 that has the same columns and hierarchy but different values:

CSV_CODIF_2.csv

it doesnt happend:

Also, if we use Tau-Argus with the first file it turns out correct:

[clara-baudry]

Sorry for the late reply. I used the files you provided and ran the analysis in Tau-Argus and in rtauargus. I got the same results in both Tau-Argus and rtauargus.

Here are the results for CSV_CODIF.csv (rtauargus in black and Tau-Argus in white) :

Here are the results for CSV_CODIF_2.csv (rtauargus in black and Tau-Argus in white) :

Thus, the problem does not seem to come from rtauargus. The package returns the results from Tau-Argus. In this case, the Tau-Argus output does not seem usable because of how far it is from the actual data. However, I would suggest using an other rounding base (I used 10 for CSV_CODIF.csv and it worked just fine). If you are determined to use a rounding base of 5, then perhaps you should post an issue on Tau-Argus.

I am also adding a comparison of the files used in Tau-Argus (_TA_2col suffix) and the ones generated by rtauargus (_5 suffix) :

Maybe you used different files to get to the last output you showed in your issue. If so could you provide us with those files so that we can try to run themon Tau-Argus and see if we get the same result.

Also, we used Tau-Argus 4.2.5 to run those tests.

I hope that this late answer is still useful.