From potthoff at uni-muenster.de Mon Jan 7 22:17:22 2002 From: potthoff at uni-muenster.de (Michael Potthoff) Date: Mon, 07 Jan 2002 22:17:22 +0100 Subject: filled histograms Message-ID: <3C3A1062.9B15F80E@uni-muenster.de> Hi statist-list! A happy new year to all of you. After i have finished my diploma-thesis on 31.12. last year i have some time to write messages. I found statist to be very useful when i worked on my project. I compiled it nearly out of the box under OS/2. I just made some minor changes to the code to make it run under OS/2. Since i use GNUPLOT 3.8 that supports filled bars i made some changes to produce filled histograms. I wanted to attach the new code, but then the mail bounces then. I'll upload the executables & source to http://hobbes.nmsu.edu soon. Meanwhile i supply the code for the filled histograms here : changes to plot.c line 263 #ifndef FILLEDHISTO fprintf(pipef, "plot [%g:%g][%i:%i] '%s' with boxes linetype linetype\n", minx-step/2., maxx+step/2., miny, maxy, GPL_DAT); #else // draw filled box fprintf(pipef, "plot [%g:%g][%i:%i] '%s' with filledboxes linetype linetype , ", minx-step/2., maxx+step/2., miny, maxy, GPL_DAT); // draw border fprintf(pipef, " '%s' with boxes linetype -1\n", GPL_DAT); #endif In order to uses this please insert : #define FILLEDHISTO to plot.h ! Regards, Michael From ccarleos at fisher.ciencias.uniovi.es Tue Jan 29 08:57:36 2002 From: ccarleos at fisher.ciencias.uniovi.es (ccarleos@fisher.ciencias.uniovi.es) Date: Tue, 29 Jan 2002 08:57:36 +0100 Subject: spanish .po? Message-ID: <20020129085736.A32176@fisher.ciencias.uniovi.es> Hello. I intend to use "statist" as a replacement to closed software we use in our classes of statistics. I think I do need a translation to Spanish of the interface, so 1. I have read the mailing list and it seems nobody has contributed a "po" file to any language. Right? 2. I have never used "gettext" before. When trying to run "xgettext" on "statist" sources, messages longer than a line (I mean, that use a backslash to escape the carriage return) cause errors. Do not know how to manage that. 3. As I begin my classes very soon, I have started to translate the messages on their place in sources -> not using gettext. I hope later, with a little Perl or a proper use of xgettext, will be easy to create the ".po" file. 4. I am obliged to use MS-DOS or Windows (will try FreeDOS) in my classes. I have compiled the 1.0.1 version of statist with DJGPP, and it works fine. But, what about international encoding in DOS? Latin 1 (iso-8859-1) corresponds with cp1252, doesn't it? Any suggestions for this job? (I have thought about using "cstocs", by Jan Kasprzak and Jan Pazdziora, to convert the sources before compiling in DOS; sorry, I did not say that I usually work in a Debian system). Thanks for any advice. And for your work. "statist" is a very useful tool in many occasions. -- _______________________________________________ Carlos Enrique Carleos Artime Departamento de Estad?stica Universidad de Oviedo Escuela de Marina Civil 33203 Spain ccarleos @ fisher.ciencias.uniovi.es http://carleos.etsiig.uniovi.es Powered by Debian GNU/Linux, http://debian.org. _______________________________________________ From bernhard at intevation.de Tue Jan 29 09:34:01 2002 From: bernhard at intevation.de (Bernhard Reiter) Date: Tue, 29 Jan 2002 09:34:01 +0100 Subject: spanish .po? In-Reply-To: <20020129085736.A32176@fisher.ciencias.uniovi.es> References: <20020129085736.A32176@fisher.ciencias.uniovi.es> Message-ID: <20020129083401.GA23845@intevation.de> Hello Carlos, On Tue, Jan 29, 2002 at 08:57:36AM +0100, ccarleos at fisher.ciencias.uniovi.es wrote: > > Hello. I intend to use "statist" as a replacement to closed software we > use in our classes of statistics. I think I do need a translation to > Spanish of the interface, so > > 1. I have read the mailing list and it seems nobody has contributed a > "po" file to any language. Right? You are right. We did not manage to use the real gettext version yet. > 2. I have never used "gettext" before. When trying to run "xgettext" on > "statist" sources, messages longer than a line (I mean, that use a > backslash to escape the carriage return) cause errors. Do not know how > to manage that. I have to look into that. It might be the case that we need to connect the lines or check if this is an error with xgettext. However I do not know when I have time to deal with this, so help is appreciated. > 3. As I begin my classes very soon, I have started to translate the > messages on their place in sources -> not using gettext. I hope later, > with a little Perl or a proper use of xgettext, will be easy to create > the ".po" file. This is fine for you purposes. Please send me a patch against 1.0.1 so I can make it available for other people. > 4. I am obliged to use MS-DOS or Windows (will try FreeDOS) in my > classes. I have compiled the 1.0.1 version of statist with DJGPP, and it > works fine. But, what about international encoding in DOS? Latin 1 > (iso-8859-1) corresponds with cp1252, doesn't it? Any suggestions for > this job? (I have thought about using "cstocs", by Jan Kasprzak and Jan > Pazdziora, to convert the sources before compiling in DOS; sorry, I did > not say that I usually work in a Debian system). It has been a long time since I've had to use DOS, so I cannot tell. Converting the sources and compiling them should get you results through trail and error. > Thanks for any advice. And for your work. "statist" is a very useful > tool in many occasions. I am happy to hear that! :) Bernhard -- Gesch?ftsf?hrer, Intevation GmbH (intevation.de) Projekt Freie GIS Software (freegis.org/index.de.html) FFII e.V. (ffii.org) FSF Europa (fsfeurope.org/index.de.html) -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 248 bytes Desc: not available Url : http://www.intevation.de/pipermail/statist-list/attachments/20020129/4b43861d/attachment.bin From carleos at airtel.net Sun Feb 17 21:50:09 2002 From: carleos at airtel.net (carleos@airtel.net) Date: Sun, 17 Feb 2002 21:50:09 +0100 Subject: spanish "translation" Message-ID: <10ae9310c29c.10c29c10ae93@airtel.net> Hi. I have done a slapdash translation of "statist" messages to Spanish. As previously announced, it's done on the sources (merely replacing deutsche expressions by spanish ones, so the default LANG=DE implies spanish), not thru a ".po" file, as would be desirable. When I learn to use "xgettext" properly, I will try again (I had problems with extraction of strings longer than a line... sure it's a problem of rtfm carefully). While waiting for that moment, I call the result "estadist". You can find in ftp://carleos.etsiig.uniovi.es/pub/statist three files: - estadist-1.0.1.diff.gz: a patch against statist-1.0.1 (this is my first patch, done with "diff -Naur"... hope it's correct) - estadist-1.0.1.tar.gz - estadist.exe: an executable for (Free)DOS systems compiled with DJGPP (http://www.delorie.com), and with sources "transliterated" thru GNU recode (http://www.gnu.org/software/recode/recode.html). I'm available at carleos @ fisher.ciencias.uniovi.es for any comment. I will tell you about my experience with "statist" in my classes. Thanks a lot. From bernhard at intevation.de Wed Feb 20 17:06:02 2002 From: bernhard at intevation.de (Bernhard Reiter) Date: Wed, 20 Feb 2002 17:06:02 +0100 Subject: Award for StatistX Message-ID: <20020220160602.GF7072@intevation.de> Most people will know about this already, but I just forgot to post it: Andreas Beyer has been awarded the "Intevation Award for Free Software". at the University of Osnabr?ck for his work on StatistX. Details: http://intevation.de/news.en.html#uni_award2001 It was also on lwn and newsforge and more. http://lwn.net/2002/0207/announce.php3 http://www.newsforge.com/article.pl?sid=02/01/31/2118232&mode=thread -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 248 bytes Desc: not available Url : http://www.intevation.de/pipermail/statist-list/attachments/20020220/2b11a80a/attachment.bin From bernhard at intevation.de Wed Feb 20 22:52:14 2002 From: bernhard at intevation.de (Bernhard Reiter) Date: Wed, 20 Feb 2002 22:52:14 +0100 Subject: spanish "translation" In-Reply-To: <10ae9310c29c.10c29c10ae93@airtel.net> References: <10ae9310c29c.10c29c10ae93@airtel.net> Message-ID: <20020220215214.GL15630@intevation.de> On Sun, Feb 17, 2002 at 09:50:09PM +0100, carleos at airtel.net wrote: > I have done a slapdash translation of "statist" messages to Spanish. Thanks for your translation. > As previously announced, it's done on the sources (merely replacing > deutsche expressions by spanish ones, so the default LANG=DE implies > spanish), not thru a ".po" file, as would be desirable. When I learn to > use "xgettext" properly, I will try again (I had problems with > extraction of strings longer than a line... sure it's a problem of rtfm > carefully). Any improvement counts. > While waiting for that moment, I call the result "estadist". > You can find in ftp://carleos.etsiig.uniovi.es/pub/statist three files: > - estadist-1.0.1.diff.gz: a patch against statist-1.0.1 (this is my > first patch, done with "diff -Naur"... hope it's correct) > - estadist-1.0.1.tar.gz > - estadist.exe: an executable for (Free)DOS systems compiled with DJGPP > (http://www.delorie.com), and with sources "transliterated" thru GNU > recode (http://www.gnu.org/software/recode/recode.html). The patch is also available from the statist pages now. The tarball had the transcoded source files and several binaries, thus I did not copy it, because it might be too confusiong for people. You might think about adding that you did the translation in the startup message. A short addition to the readme in the tarball probably also would be appropriate. You should get the credit of course and people otherwise might contact me and assume that I speak Spanish. :) > I'm available at carleos @ fisher.ciencias.uniovi.es for any comment. > I will tell you about my experience with "statist" in my classes. I am really interested in your experiences. Also let me know if you compare statist to other results as other programs or manual calculations. Cheers, Bernhard -- Professional Service for Free Software (intevation.net) The FreeGIS Project (freegis.org) Association for a Free Informational Infrastructure (ffii.org) FSF Europe (fsfeurope.org) -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 248 bytes Desc: not available Url : http://www.intevation.de/pipermail/statist-list/attachments/20020220/ac85764e/attachment.bin From carleos at fisher.ciencias.uniovi.es Mon Mar 11 13:12:39 2002 From: carleos at fisher.ciencias.uniovi.es (Carleos Artime Fi) Date: Mon, 11 Mar 2002 13:12:39 +0100 Subject: bug in Makefile? Message-ID: <20020311131239.A12554@fisher.ciencias.uniovi.es> Hi. In statist-1.0.1, tests/Makefile, this line appears: test -z "$(CLEANFILES)" || rm $(CLEANFILES) which, in my Debian 2.2r5 produces an error, because "test" returns 1 if true and 0 if false, and "bash" executes the command after || if success (0) and does not if failed (not 0). Am I wrong? On the other hand, I would like to know if there has been any attempt to build a Debian package of "statist" (a .deb file). Anybody knows? Cheers. Carlos Carleos From bernhard at intevation.de Mon Mar 11 19:21:00 2002 From: bernhard at intevation.de (Bernhard Reiter) Date: Mon, 11 Mar 2002 19:21:00 +0100 Subject: bug in Makefile? In-Reply-To: <20020311131239.A12554@fisher.ciencias.uniovi.es> References: <20020311131239.A12554@fisher.ciencias.uniovi.es> Message-ID: <20020311182100.GA7561@intevation.de> On Mon, Mar 11, 2002 at 01:12:39PM +0100, Carleos Artime Fi wrote: > Hi. In statist-1.0.1, tests/Makefile, this line appears: > > test -z "$(CLEANFILES)" || rm $(CLEANFILES) > > which, in my Debian 2.2r5 produces an error, because "test" returns 1 if > true and 0 if false, and "bash" executes the command after || if > success (0) and does not if failed (not 0). The rm will be executed when the "$(CLEANFILES)" is not empty. This seem to be correct to me. In my bash I can try the following: test -z hi || echo hi hi test -z || echo ho This is because test -z returns 0 if the string is empty. Hmm no idea why this would produce an error. > On the other hand, I would like to know if there has been any attempt to > build a Debian package of "statist" (a .deb file). Anybody knows? I am not aware of debian packaging attempts. It would be appreciated, though. :) Bernhard -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 248 bytes Desc: not available Url : http://www.intevation.de/pipermail/statist-list/attachments/20020311/1ffbd5ed/attachment.bin From carleos at fisher.ciencias.uniovi.es Tue Mar 12 21:42:50 2002 From: carleos at fisher.ciencias.uniovi.es (Carleos Artime Fi) Date: Tue, 12 Mar 2002 21:42:50 +0100 Subject: bug in Makefile? In-Reply-To: <"from bernhard"@intevation.de> References: <20020311131239.A12554@fisher.ciencias.uniovi.es> <20020311182100.GA7561@intevation.de> Message-ID: <20020312214250.A25397@fisher.ciencias.uniovi.es> > This is because test -z returns 0 if the string is empty. > Hmm no idea why this would produce an error. Sorry! I misunderstood the behavior of "test". The problem must be some conflict between statist's Makefile and Debian tools for package building. I will try anew and tell you the results. carlos From sfronzek at gmx.net Mon Apr 22 16:42:32 2002 From: sfronzek at gmx.net (sfronzek@gmx.net) Date: Mon, 22 Apr 2002 16:42:32 +0200 (MEST) Subject: Example shell-script for using STATIST in batch-mode Message-ID: <6985.1019486552@www35.gmx.net> Hi statist-list, I have started using the batch-mode functionality of STATIST to calculate simple statistical numbers such as the standard deviation for a large number of files. Maybe some of you find the shell-script below useful so I post it to this list. The task of it is to extract some data from a relatively large file and calculate the standard deviation for it. The script is by no means a masterpiece of shell-programming, but I hope to give an example of how STATIST can be used effectively in batch-mode. Examples of the data files for which I made the shell-script can be found from http://www.knmi.nl/samenw/eca/htmls/index3.html. Copy the script below to std.sh and the batch-commands to a file STAT_batch.txt (note the empty lines!). Best regards, Stefan. >>>>>>>>BEGINNING OF FILE std.sh<<<<<<<<<< #! /usr/bin/sh # std.sh # Calculates the standard deviation for each month of every year # from a table with daily values by calling STATIST in batch-mode. # The data file is expected to be comma-separated table with one line per # day, the date denoted as yyyymmdd in the second column, and the data for # which to calculate the standard deviation in forth column. The output # goes to standard out as well as separate files for each month. echo std.sh: Processing file $1 # list of months for which to do the calculations monthlist="01 02 03 04 05 06 07 08 09 10 11 12"; for month in $monthlist ; do # first year to calculate year=1951; # set the file name for output outfile=$1$month.dat; # loop through the years until 1980 while [ $year -lt 1980 ]; do # grep out the values for calculation: all days of one month in one year grep $year$month $1 | \ # use the 4th column = mean temperature cut -d, -f4 | \ # substitute the missing-value denotion -9999 with the STATIST denotion 'M' sed '1,$ s/-9999/ M/' > tmp.dat; # call STATIST in batch-mode to calculate the standard deviation stddev=$(statist tmp.dat -silent -noplot < STAT_batch.txt | grep "Standard deviation" | cut -d: -f2); # print out: first to standard out, then to separate files for each month echo $year$month $stddev; echo $year$month $stddev >> $outfile; # increment the year counter year=`expr $year + 1`; done done # remove the tmp file rm tmp.dat; >>>>>>>>END OF FILE std.sh<<<<<<<<<< >>>>>>>>START OF FILE STAT_batch.txt<<<<<<<<<< 4 1 a 0 >>>>>>>>END OF FILE STAT_batch.txt<<<<<<<<<< -- GMX - Die Kommunikationsplattform im Internet. http://www.gmx.net From bernhard at intevation.de Tue Apr 30 16:13:55 2002 From: bernhard at intevation.de (Bernhard Reiter) Date: Tue, 30 Apr 2002 16:13:55 +0200 Subject: SalStat: Knowing StatistX? Message-ID: <20020430141355.GH14990@intevation.de> Alan, just saw your wxPython statistic application. http://salstat.sunsite.dk/ It looks quite promissing, I hope to find time to try it. In your analysis you write: Why another statistics package? GNU/Linux doesn't have many free statistics packages. The most capable ones rely upon the user learning a command line interface (except for the lisp-stat add-on called Vista). A few years ago I did a similiar analysis about statistic application which are Free Software. It is great to see that you've came to similiar conclusions and started a project with very reasonable technical decisions regarding GUI toolkit (WxWindows) and programming language (Python). Please note that your list of Free Software GUI statistic applications is missing "StatistX" based on "Statist". http://www.usf.uni-osnabrueck.de/~abeyer/private/StatistX/ http://www.usf.uos.de/~breiter/tools/statist/index.de.html Statist and StatistX started a lot earlier, but are quite useful and tested in their areas. A link to the category in SAL might be useful the readers of your page to find other Free Software dealing with statistics. http://sal.kachinatech.com/A/3/ Best Regards, Bernhard -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 248 bytes Desc: not available Url : http://www.intevation.de/pipermail/statist-list/attachments/20020430/998e7582/attachment.bin From carleos at fisher.ciencias.uniovi.es Thu May 16 20:15:36 2002 From: carleos at fisher.ciencias.uniovi.es (Carleos Artime Fi) Date: Thu, 16 May 2002 20:15:36 +0200 Subject: debian packages In-Reply-To: <"from bernhard"@intevation.de> References: <20020311131239.A12554@fisher.ciencias.uniovi.es> <20020311182100.GA7561@intevation.de> Message-ID: <20020516201536.A9013@fisher.ciencias.uniovi.es> Hi! This is my first attempt to build Debian packages of "statist". You may access this anonymous ftp server: ftp://fisher.ciencias.uniovi.es/pub/statist and try it out (i386 binaries). If no bugs are found, my intention is that statist be included in the main section of Debian in a near future... All the best. carlos PS: For any problem, please e-mail carleos @ fisher.ciencias.uniovi.es. From bernhard at intevation.de Fri May 17 11:18:35 2002 From: bernhard at intevation.de (Bernhard Reiter) Date: Fri, 17 May 2002 11:18:35 +0200 Subject: debian packages In-Reply-To: <20020516201536.A9013@fisher.ciencias.uniovi.es> References: <20020311131239.A12554@fisher.ciencias.uniovi.es> <20020311182100.GA7561@intevation.de> <20020516201536.A9013@fisher.ciencias.uniovi.es> Message-ID: <20020517091835.GD4995@intevation.de> Carleos, this is good news! Can you also make the .src.deb available somewhere? The good part about Debian is, that it runs on many hardware platforms. :) About the language issue, wouldn't it be more suitable to give to give the packages different names. Especially if we manage to do a gettext version sometime in the future? E.G. like: statist-de-.. statist-en-.. Thanks for sharing your work, Bernhard On Thu, May 16, 2002 at 08:15:36PM +0200, Carleos Artime Fi wrote: > This is my first attempt to build Debian packages of "statist". > You may access this anonymous ftp server: > ftp://fisher.ciencias.uniovi.es/pub/statist > and try it out (i386 binaries). > > If no bugs are found, my intention is that statist be included > in the main section of Debian in a near future... -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 248 bytes Desc: not available Url : http://www.intevation.de/pipermail/statist-list/attachments/20020517/6f4164a4/attachment.bin From bernhard at intevation.de Mon May 20 15:37:48 2002 From: bernhard at intevation.de (Bernhard Reiter) Date: Mon, 20 May 2002 15:37:48 +0200 Subject: statist question In-Reply-To: References: Message-ID: <20020520133748.GD994@intevation.de> Hello Jonne, [Answer also forwarded to the archived statist mailinglist.] In the original documentation (written by Dirk Melcher in German) there is an explained figure of the box and whisker plot. It says that the '[' and ']' characters stand for the "90% confidence intervall of the median". The function was written by Dirk. He quotes a German book on statistics. I've not rechecked the calculation which happens in the lines you identified. On Sun, May 19, 2002 at 06:50:11PM +0200, J. Zutt wrote: > Could you please explain me what the '[' and ']' characters > refer to in a box-whisker plot? > I looked in the source code and found this: > no_u = median + 1.58*(q_u-q_l)/sqrt((REAL)n); > no_l = median - 1.58*(q_u-q_l)/sqrt((REAL)n); > > I thought it has something todo with it, but I cannot > figure it out. > Furthermore, is it possible to plot more than one > box-whisker plot (when 'column for variable 1' is > asked, I cannot enter "all"...) in one figure? You cannot use the function to plot more than one box and whisker plot at once. If you need several plots you could write a small script. There is one trick to get open several gnuplot windows. You have to enter raw gnuplot commands and tell gnuplot to send the next plot in a different X11 window, like: set term x11 1 For a script you can also make sure that gnuplot places the plot result in a postscript file: set term postscript eps or set term postscript you can even give this file a name an put several plot in there: set output "result1.eps" Cheers, Bernhard -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 248 bytes Desc: not available Url : http://www.intevation.de/pipermail/statist-list/attachments/20020520/148a068d/attachment.bin From abeyer at usf.uni-osnabrueck.de Fri Jun 7 13:46:17 2002 From: abeyer at usf.uni-osnabrueck.de (Andreas Beyer) Date: Fri, 07 Jun 2002 13:46:17 +0200 Subject: StatistX.org Message-ID: <3D009D09.901A2B79@usf.uni-osnabrueck.de> Hi, StatistX now has its own internet domain: www.statistx.org This is easier to remember. Andreas -- ----------------------------------------------- Andreas Beyer - abeyer at uos.de http://www.usf.uos.de/~abeyer/ phone: +49-541-969 2573 fax: -969 2599 ----------------------------------------------- From abeyer at usf.uni-osnabrueck.de Fri Jun 7 13:58:06 2002 From: abeyer at usf.uni-osnabrueck.de (Andreas Beyer) Date: Fri, 07 Jun 2002 13:58:06 +0200 Subject: statist + XML Message-ID: <3D009FCE.1AFE6720@usf.uni-osnabrueck.de> Hi, I am working on an XML-interface for statist. This e-mail is only to avoid double work. I cannot present results, yet. Thus, if you intend to do the same, contact me and we can pool our resources. I suggest, that XML becomes statist's native data format. I will develop a converter that converts old statist data files into XML files. The converter will work as a stand alone program and it can be incorporated into statist. The DTD will be very simple. It only has two basic elements: 'table' having the attribute 'title' 'column' having the attributes 'title' and 'type' 'type' can be either 'string' or 'numerical' This definition is compatible to many other DTD and extendable (e.g. another 'type' could be 'bin' or anything else). Row lables (StatistX only): If the first column is of type 'string' it will be interpreted as labels of the rows. Andreas -- ----------------------------------------------- Andreas Beyer - abeyer at uos.de http://www.usf.uos.de/~abeyer/ phone: +49-541-969 2573 fax: -969 2599 ----------------------------------------------- From elmar at informatik.Uni-Osnabrueck.DE Mon Jun 10 17:19:07 2002 From: elmar at informatik.Uni-Osnabrueck.DE (Elmar Ludwig) Date: Mon, 10 Jun 2002 17:19:07 +0200 Subject: statist + XML In-Reply-To: <3D009FCE.1AFE6720@usf.uni-osnabrueck.de> Message-ID: <68017770-7C85-11D6-A24C-0050E41EBA54@informatik.uni-osnabrueck.de> Andreas Beyer wrote: > I am working on an XML-interface for statist. Why XML? > The DTD will be very simple. It only has two basic elements: > 'table' having the attribute 'title' > 'column' having the attributes 'title' and 'type' > 'type' can be either 'string' or 'numerical' I can't imagine how this is supposed to look like. Don't you need an element for the cell contents? > This definition is compatible to many other DTD and extendable > (e.g. another 'type' could be 'bin' or anything else). Do you have a DTD already (to look at)? Elmar. -- Elmar Ludwig elmar at informatik.uni-osnabrueck.de Dept. of Math. & Computer Science +49 541-969-2491 (Tel) University of Osnabrueck +49 541-969-2799 (Fax) 49069 Osnabrueck, Germany http://www.inf.uos.de/elmar/