search

sdcTools / UserSupport

The place to be for User Support on SDC tools and to download the latest releases
https://sdctools.github.io/UserSupport/
Other
11 stars 3 forks source link

format of record keys #105

Open r-tent opened 5 years ago

r-tent commented 5 years ago

Please specify
SDC tool used: tau-argus, sdcTable Version used: 4.1.12BETA Operating system used: Windows

Hello @ppdewolf,

sorry for getting on your nerves again, but I have one further remark of something that took me a while, when testing the cell key method of tau-Argus. Even though it was more a problem of myself than of your program, maybe you could fix the following "issue".

When I light heartedly added record keys to my data via R , I didn't notice that some values where stored in a form like 6e-05 instead of 0.00006. Now for tau-Argus, this was quite a big deal while for me it was rather untraceable, since some csv-viewiers convert this expression directly, when opening the file. I think it would be great if tau-Argus accepted values like 6e-05 for record keys :-)

ppdewolf commented 5 years ago

Good point. Something to add in a newer version or at least something to mention in the quick reference and manual.

ppdewolf commented 5 years ago

Are you using fixed format or free format input? I couldn't reproduce the error in fixed format.

Do you have the same problem when using a response variable in that format (i.e., e.g., turnover 6.7861e4)?

ppdewolf commented 5 years ago

This seems to occur only with free format. Happens with any numeric variable (both very small and very large numbers).

Work around for record keys I've updated the example how to produce the record keys in R such that the scientific notation is not used and the resulting numbers all have 15 decimals. Something like this: # Record keys (direct approach) microdata <- read.csv2(file=”hc_9_2_synth.csv”) set.seed(123) microData$recordkey <- format(runif(dim(microData)[1]), scientific=FALSE, nsmall=15) write.table(microData, ”hc_9_2_synth_Tau.csv”, row.names=FALSE, col.names=FALSE, quote=FALSE, dec=”.”,sep=”;”)