From e138052b94a547c11ae36c79cb91a7b163cd4105 Mon Sep 17 00:00:00 2001
From: j3nnn1 <jcmm986@gmail.com>
Date: Tue, 29 May 2012 01:43:33 -0300
Subject: [PATCH] script python to blanked one field on csv

---
 weka/putblankfield001.py | 47 ++++++++++++++++++++++++++++++++++++++++
 weka/treeexercise1.sh    | 44 +++++++++++++++++++++++++++++++++++++
 2 files changed, 91 insertions(+)
 create mode 100644 weka/putblankfield001.py
 create mode 100755 weka/treeexercise1.sh

diff --git a/weka/putblankfield001.py b/weka/putblankfield001.py
new file mode 100644
index 0000000..4fe7e68
--- /dev/null
+++ b/weka/putblankfield001.py
@@ -0,0 +1,47 @@
+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+# j3nnn1 - 0.0.0 -
+
+import os
+import string
+import random
+
+
+def cleanstring(l):
+    l[-1] = l[-1].rstrip("\n")
+    return([x.strip() for x in l])
+
+def getpercenta(percent, numberlines):
+    result = percent * numberlines
+    return result
+    
+f = open("outputtree1.csv", "r")
+output = open("output.csv", "w")
+numlines = 0
+
+for line in f.readlines():
+    numlines = numlines + 1
+
+endlines = int(getpercenta(0.25, numlines))
+
+f.seek(0)
+line = f.readline()
+i = 0
+j = 0
+while line:
+    i = i + 1
+    a = line.split(",")
+    aux = cleanstring(a)
+    
+    if i%2 and j<=endlines:
+        j = j +1 
+        indice = random.randint(0, (len(aux)-1))
+        aux[indice] = ''
+
+    aux[-1] = aux[-1]+"\n"
+    newline = ', '.join(aux)
+    output.write(newline)
+    line = f.readline()
+
+f.close()
+output.close()
diff --git a/weka/treeexercise1.sh b/weka/treeexercise1.sh
new file mode 100755
index 0000000..4f97b26
--- /dev/null
+++ b/weka/treeexercise1.sh
@@ -0,0 +1,44 @@
+#!/bin/bash
+#	#bash_treedecision.sh 
+#	#Path to weka.jar, received a path absolute 
+#	#File to process, received a path absolute
+#	#classpath 
+#	#interval confidence factor
+#	#Options
+#	#args=$#
+pathtoweka=$1
+file=$2
+classpath=$3
+options=$5
+
+if [ -z  $file ]; then
+echo "usage ./bash_treedecision.sh <pathtoweka> <filetoprocess> <options>"
+exit 1
+fi
+
+if [ -z $options ]; then
+options="-i -M 2 -no-cv"
+fi
+confidencefactor="0.0"
+cont=25
+
+echo "size_tree,leaves,confianza,percen_correct,cant_correct"
+while [ $cont -le 501 ]; do 
+	if [ $cont -gt 75 ]; then
+		confidencefactor="0."
+	fi
+	#echo "java -classpath $pathtoweka weka.classifiers.trees.J48 -C $confidencefactor$cont $options -t $file"
+   	size=`java -classpath $pathtoweka weka.classifiers.trees.J48 -C $confidencefactor$cont $options -t $file | grep "Size"`;
+   	leaves=`java -classpath $pathtoweka weka.classifiers.trees.J48 -C $confidencefactor$cont $options -t $file | grep "Leaves"`;
+	correctly=`java -classpath $pathtoweka weka.classifiers.trees.J48 -C $confidencefactor$cont $options -t $file | grep "Correctly Classified Instances"`;
+	int_size="$(echo $size | awk '{print $6}')"
+	int_leaves="$(echo $leaves | awk '{print $5}')"
+	int_cases="$(echo $correctly | awk '{print $4}')"
+	prc_cases="$(echo $correctly | awk '{print $5}')"
+	echo "$int_size, $int_leaves, $confidencefactor$cont, $prc_cases, $int_cases"	
+   	cont=`expr $cont + 25`
+done
+exit 0
+
+
+