Htydjtk

See5 / C5.0 is Data Mining Tools available from rulequest

I want to compile C50 for Linux, preferably for CentOS 6.x, but I am unable to compile. I have also tried on Ubuntu, but not success there as well.

I have downloaded C50.tgz from C5.0 Release 2.07 GPL Edition
After extracting when I run ./Makefile it gives below error on Ubuntu18

./Makefile: line 9: CC: command not found

./Makefile: line 10: CFLAGS: command not found

./Makefile: line 11: S: command not found

./Makefile: line 11: LFLAGS: command not found

./Makefile: line 12: SHELL: command not found

./Makefile: line 19: src: command not found

./Makefile: line 48: obj: command not found

./Makefile: line 59: all:: command not found

cat defns.i global.c c50.c construct.c formtree.c info.c discr.c contin.c subset.c prune.c p-thresh.c trees.c siftrules.c ruletree.c rules.c getdata.c implicitatt.c mcost.c confmat.c sort.c update.c attwinnow.c classify.c formrules.c getnames.c modelfiles.c utility.c xval.c

        | egrep -v 'defns.i|extern.i' >c50gt.c

make: /bin/csh: Command not found

Makefile:75: recipe for target 'c5.0' failed

make: *** [c5.0] Error 127

./Makefile: line 61: CC: command not found

./Makefile: line 61: LFLAGS: command not found

./Makefile: line 61: -o: command not found

./Makefile: line 67: obj: command not found

./Makefile: line 67: c5.0dbg:: command not found

./Makefile: line 68: CC: command not found

./Makefile: line 68: obj: command not found

./Makefile: line 68: -g: command not found

./Makefile: line 74: src: command not found

./Makefile: line 74: c5.0:: command not found

./Makefile: line 75: src: command not found

./Makefile: line 77: CC: command not found

./Makefile: line 77: LFLAGS: command not found

./Makefile: line 77: -O3: command not found

strip: 'c5.0': No such file

./Makefile: line 82: obj: command not found

./Makefile: line 85: .c.o:: command not found

./Makefile: line 86: syntax error near unexpected token `newline'

./Makefile: line 86: `  $(CC) $(CFLAGS) -c $<'

If I run make command, it gives below error.

make c5.0 

make: /bin/csh: Command not found Makefile:60: recipe for

target 'all' failed make: *** [all] Error 127

gcc version is:

gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0

After extracting C50.tgz file, give execution rights to Makefile.

chmod +x Makefile

Then install csh

sudo apt install csh

Run below command to check if csh is install and check csh version (if installed)

dpkg -l csh

It should show output like below.

Desired=Unknown/Install/Remove/Purge/Hold

| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend

|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)

||/ Name                           Version              Architecture         Description

+++-==============================-====================-====================-==================================================================

ii  csh                            20110502-3           amd64                Shell with C-like syntax

Now run below command.

./Makefile

It should give output like below. (Ignore the warning messages).

./Makefile: line 9: CC: command not found

./Makefile: line 10: CFLAGS: command not found

./Makefile: line 11: S: command not found

./Makefile: line 11: LFLAGS: command not found

./Makefile: line 12: SHELL: command not found

./Makefile: line 19: src: command not found

./Makefile: line 48: obj: command not found

./Makefile: line 59: all:: command not found

cat defns.i global.c c50.c construct.c formtree.c info.c discr.c contin.c subset.c prune.c p-thresh.c trees.c siftrules.c ruletree.c rules.c g     etdata.c implicitatt.c mcost.c confmat.c sort.c update.c attwinnow.c classify.c formrules.c getnames.c modelfiles.c utility.c xval.c

        | egrep -v 'defns.i|extern.i' >c50gt.c

gcc -ffloat-store  -O3 -o c5.0 c50gt.c -lm

c50gt.c: In function ‘ListAttsUsed’:

c50gt.c:14025:12: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]

      Att = (Attribute) DefSVal(D[e]);

            ^

c50gt.c: In function ‘Error’:

c50gt.c:15561:17: warning: format not a string literal and no format arguments [-Wformat-security]

     fprintf(Of, Buffer);

                 ^~~~~~

strip c5.0

rm c50gt.c

./Makefile: line 61: CC: command not found

./Makefile: line 61: LFLAGS: command not found

./Makefile: line 61: -o: command not found

./Makefile: line 67: obj: command not found

./Makefile: line 67: c5.0dbg:: command not found

./Makefile: line 68: CC: command not found

./Makefile: line 68: obj: command not found

./Makefile: line 68: -g: command not found

./Makefile: line 74: src: command not found

./Makefile: line 74: c5.0:: command not found

./Makefile: line 75: src: command not found

./Makefile: line 77: CC: command not found

./Makefile: line 77: LFLAGS: command not found

./Makefile: line 77: -O3: command not found

./Makefile: line 82: obj: command not found

./Makefile: line 85: .c.o:: command not found

./Makefile: line 86: syntax error near unexpected token `newline'

./Makefile: line 86: `  $(CC) $(CFLAGS) -c $<'

You will see a new file c5.0*
Check command line options with below command.

./c5.0 -h

it should show below output.

C5.0 [Release 2.07 GPL Edition]         Fri Mar 15 17:43:39 2019

-------------------------------



    Options:

        -f <filestem>   application filestem

        -r              use rule-based classifiers

        -u <bands>      order rules by utility in bands

        -w              invoke attribute winnowing

        -b              invoke boosting

        -t <trials>     number of boosting trials

        -p              use soft thresholds

        -e              focus on errors (ignore costs file)

        -s              find subset tests for discrete atts

        -g              do not use global tree pruning

        -m <cases>      restrict allowable splits

        -c <percent>    confidence level (CF) for pruning

        -S <percent>    training sample percentage

        -X <folds>      cross-validate

        -I <integer>    random seed for sampling and cross-validation

        -h              print this message

Now C50 is ready to be used. Here is a sample command(assuming sampledata.data and sampledata.names files exists in the same folder).

./c5.0 -r -f sampledata

After successful execution of the command, sampledata.rules will be generated in the same folder.

See full tutorial from here