Newbie: Starting RapidMiner from Command Line

IngoRM

Original message from SourceForge forum at http://sourceforge.net/forum/forum.php?thread_id=2042410&;forum_id=390413

I need to start RapidMiner from the command line, but I can't get it to load and run the referenced XML file. For example, if I type "rapidminer 03_XValidation_Numerical.xml", IAW with the documentation, the GUI loads but the nothing else happens. When I check the xml it is empty, i.e.

<operator name="Root" class="Process" expanded="yes">
</operator>

I have "RapidMiner" installed in the "C:\Program Files\Rapid-I\RapidMiner-4.1" dir and the sample files installed in the "C:\Program Files\Rapid-I\RapidMiner-4.1\sample" dir. I'm using XP and I updated the path. Any ideas on what I need type on the command line to get the xml file to run? 

Thank you in advance...


Answer by Ingo:

Hi,

I just tested this myself on Windows XP by going into the scripts directory and typing

rapidminer ../sample/04_Validation/03_XValidation_Numerical.xml

and it works. The output looks like this:

C:\home\ingo\workspace\yale\scripts>rapidminer ../sample/04_Validation/03_XVali
ation_Numerical.xml
RAPIDMINER_HOME is not set. Trying the directory 'C:\home\ingo\workspace\yale\s
ripts\..'...
Using local jre: C:\home\ingo\workspace\yale\scripts\..\jre\bin\java.exe...
Starting RapidMiner from 'C:\home\ingo\workspace\yale\scripts\..' using classes
from 'C:\home\ingo\workspace\yale\scripts\..\build'...
RapidMiner version 4.1, Copyright (C) 2001-2007
RapidMiner comes with ABSOLUTELY NO WARRANTY; This is free software,
and you are welcome to redistribute it under certain conditions;
see license information in the file named LICENSE.
G 18.05.2008 16:52:14: rapidminer.home is 'C:\home\ingo\workspace\yale\scripts\
.'.
Last message repeated 1 times.
G May 18, 2008 4:52:14 PM: ----------------------------------------------------
G May 18, 2008 4:52:14 PM: Initialization Settings
G May 18, 2008 4:52:14 PM: ----------------------------------------------------
G May 18, 2008 4:52:14 PM: Default system encoding for IO: windows-1252
G May 18, 2008 4:52:14 PM: Load core operators...
G May 18, 2008 4:52:14 PM: Load Weka operators: true
G May 18, 2008 4:52:14 PM: Load JDBC drivers from lib directory: true
G May 18, 2008 4:52:14 PM: Load JDBC drivers from classpath: false
G May 18, 2008 4:52:14 PM: Load plugins: true
G May 18, 2008 4:52:14 PM: Load plugins from 'C:\home\ingo\workspace\yale\scrip
s\..\lib\plugins'
G May 18, 2008 4:52:14 PM: ----------------------------------------------------
G May 18, 2008 4:52:14 PM: Read rcfile 'C:\home\ingo\workspace\yale\scripts\..\
tc\rapidminerrc'.
G May 18, 2008 4:52:14 PM: Trying rcfile 'C:\home\ingo\workspace\yale\scripts\.
\etc\rapidminerrc.Windows XP'...skipped
G May 18, 2008 4:52:14 PM: Trying rcfile 'C:\Dokumente und Einstellungen\ingo\.
apidminer\4_1_0_rapidminerrc'...skipped
G May 18, 2008 4:52:14 PM: Read rcfile 'C:\Dokumente und Einstellungen\ingo\.ra
idminer\4_1_0_rapidminerrc.Windows XP'.
G May 18, 2008 4:52:14 PM: Trying rcfile 'C:\home\ingo\workspace\yale\scripts\r
pidminerrc'...skipped
G May 18, 2008 4:52:14 PM: Trying rcfile 'C:\home\ingo\workspace\yale\scripts\r
pidminerrc.Windows XP'...skipped
G May 18, 2008 4:52:14 PM: Trying rapidminer.rcfile. Property not specified...s
ipped
G May 18, 2008 4:52:14 PM: Loading operators from 'operators.xml'.
G May 18, 2008 4:52:26 PM: Loading JDBC driver information from 'etc:jdbc_prope
ties.xml'.
P May 18, 2008 4:52:26 PM: Initialising process setup
P May 18, 2008 4:52:26 PM: [NOTE] No filename given for result file, using stdo
t for logging results!
P May 18, 2008 4:52:26 PM: Checking properties...
P May 18, 2008 4:52:26 PM: Properties are ok.
P May 18, 2008 4:52:26 PM: Checking process setup...
P May 18, 2008 4:52:26 PM: Inner operators are ok.
P May 18, 2008 4:52:26 PM: Checking i/o classes...
P May 18, 2008 4:52:26 PM: i/o classes are ok. Process output: PerformanceVecto
.
P May 18, 2008 4:52:26 PM: Process ok.
P May 18, 2008 4:52:26 PM: Process initialised
P May 18, 2008 4:52:26 PM: [NOTE] Process starts
P May 18, 2008 4:52:26 PM: Process:
Root[0] (Process)
+- Input[0] (ExampleSource)
+- XVal[0] (XValidation)
+- Training[0] (LibSVMLearner)
+- ApplierChain[0] (OperatorChain)
+- Test[0] (ModelApplier)
+- Evaluation[0] (RegressionPerformance)
P May 18, 2008 4:52:36 PM: [NOTE] Process finished after 10 seconds
P May 18, 2008 4:52:36 PM: Process:
Root[1] (Process)
+- Input[1] (ExampleSource)
+- XVal[1] (XValidation)
+- Training[10] (LibSVMLearner)
+- ApplierChain[10] (OperatorChain)
+- Test[10] (ModelApplier)
+- Evaluation[10] (RegressionPerformance)
P May 18, 2008 4:52:36 PM: Produced output:
IOContainer (1 objects):

PerformanceVector [
-----root_mean_squared_error: 7.271 +/- 1.398 (mikro: 7.404 +/- 0.000)
-----absolute_error: 5.107 +/- 0.949 (mikro: 5.107 +/- 5.361)
-----relative_error: 0.426 +/- 0.364 (mikro: 0.426 +/- 2.210)
-----normalized_absolute_error: 0.040 +/- 0.014 (mikro: 0.038)
-----root_relative_squared_error: 0.044 +/- 0.014 (mikro: 0.041)
-----squared_error: 54.826 +/- 20.284 (mikro: 54.826 +/- 129.593)
-----correlation: 0.999 +/- 0.001 (mikro: 0.999)
]
(created by Evaluation)

P May 18, 2008 4:52:36 PM: [NOTE] Process finished successfully
Process finished successfully


I am hence assuming there is something wrong with your path settings...

Cheers,
Ingo


Answer by topic starter:

Thank you very much for your prompt response and taking the time to answer such a newbie question. After reading your post I was able run Rapid Miner from the command line and it now works perfectly from the scripts directory. However, after updating my path settings, if I try to run Rapid Miner on Windows XP using the "START" >> "RUN" function, I still need to explicitly write the path, i.e. 

"C:\Program Files\Rapid-I\RapidMiner-4.1\scripts\rapidminer" ..\sample\04_Validation\03_XValidation_Numerical.xml

Although I still can't rule out a problem with my path settings (e.g. "C:\Program Files\Rapid-I\RapidMiner-4.1" & "C:\Program Files\Rapid-I\RapidMiner-4.1\scripts"), but thats just a minor issue since the program works fine.

Thanks again for your great support...


Answer by Steffen:

Hello

just a suggestion:
"updating my path settings" => I guess you refer to the path in system->environment etc.
As far as I know,the windows path dont like blanks => (Program Files)

greetings

Steffen

wessel

Is this also supposed to work on linux?

I get:
wessel@wegumar7:~/Desktop/rapidminer/scripts$ ls
rapidminer  rapidminer.bat  RapidMinerGUI  RapidMinerGUI.bat
wessel@wegumar7:~/Desktop/rapidminer/scripts$ chmod +x rapidminer
wessel@wegumar7:~/Desktop/rapidminer/scripts$ ./rapidminer
RAPIDMINER_HOME is not set. Trying the directory '.'...
No maximum Java memory defined, using 512 Mb...
Neither ./build nor ./lib/rapidminer.jar was found.
Try 'ant build' or 'ant dist' first or check the value of RAPIDMINER_HOME!

colo

This works fine for me. But in my case the parent folder is searched for RapidMiner home directory instead of the current folder as in your case (RAPIDMINER_HOME is not set. Trying the directory '.'...). Did you try setting the RAPIDMINER_HOME variable to ~/Desktop/rapidminer/ before the execution of the script?
On windows you should execute rapidminer.bat, or do you use some sort of Linux shell emulation?

Regards,
Matthias

wessel

My bad, I wrote windows instead of Linux.
The question was if this also works on Linux.

Marin

Is it possible to execute RapidMiner processes without using the graphical interface?
E.g.: I type in command line something and after few days in output file/directory I get the results?

If I can do it, please let me know how to (I have the possibility to use a grid computer and I am thinking about running few serial processes on it as long as it takes).
Suppose I have process named Process.xml, input named whateve.xyz and output named cool_results.zyx, and all other files I have to have. What has to be installed on the grid/cluster in order for this to work?
Some steps or reference to RMwiki, Manual for "console dummies" or example would be helpful.

Cheerz,

Marin

wessel

What has to be installed on the grid/cluster in order for this to work?

I think you need to following steps:
Go the the rapid-miner folder and use:
- ant build
- chmod 700 scripts/rapidminer
- export PATH=${PATH}:`pwd`/scripts
- export RAPIDMINER_HOME=`pwd`

Then you can run rapidminer from anywhere using (for example):
rapidminer com.rapidminer.
RapidMinerCommandLine -f MyProcess.rmp

Note that ` is not your ordinary ' but its the one next to 1 on the keyboard.

Here is the windows version I got from Cristiano:
http://www.rivolinet.it/wordpress/2011/04/28/run-rapidminer-5-from-the-command-line-windows/
He notes that it is important to unselect the ‘Resolve relative…’, if you want to read files from your repository.

#############################

This said, this process is really cumbersome.
I am not allowed to install Rapid Analytics on my University Grid.
So this is not an option.
Furthermore Rapid Analytic is not able to distribute a processes amongst many computers.

In many cases, where you want to train lots of models, this process is almost trivial to run in parallel.
I managed to a 1000+ times speedup using my own Java code,
Given different data to different algorithms on a single node in the grid distributed over different threads:

static void execute() {
ExecutorService es = Executors.newFixedThreadPool(THREADS);
for (int start = BEGIN; start < END; start += K) {
int stop = start + K;
if (stop > END) {
stop = END;
}
es.execute(alg(start, stop));
}
es.shutdown();
}

And using a bash script to call this Java Class many times with different parameters,
which automatically assigns workload to different nodes on the grid.

TheBear

hello, I tried to run rm5 under windows7 via commandline. Unfortunately it failed with the following message

11.08.2011 14:36:23 com.rapidminer.RapidMinerCommandLine run SEVERE: Cannot read process setup 'c:\users\XXX\desktop\test.xml': No absolute path: c:\users\XXX\desktop\test.xml

I started with
"C:\Program Files\Rapid-I\RapidMiner5\scripts\rapidminer" c:\users\XXX\desktop\test.xml

Anyone who might help?
Thx
Sebastian

Marco_Boeck

Hi,

if possible you should use "//RepositoryName/ProcessName" as your argument, because RapidMiner works with repositories rather than the actual file system.
However, you can still use this:
-f "c:\users\XXX\desktop\test.rmp"


Regards,
Marco