Reverted library path, added more instructions

linearclassification/README.md

@@ -22,7 +22,7 @@ You specify how much data to hold back for validation on the command line using
 
 ##Dependencies
 
-* python v2.7 or greater
+* Python v2.7 or greater
 * scikit-learn: http://scikit-learn.org/stable/install.html
 
 
@@ -65,9 +65,15 @@ You can create your own configuration by inheriting from the ConfigBase module (
 
 
 
-##Command line
+##Executing
 
-You execute like this:
+Firstly, clone the repository to your local machine.
+
+Next, you'll need to add the library to your PYTHONPATH environment setting by modifying your .profile or .bash_profile file, for example:
+
+<pre>export PYTHONPATH=$PYTHONPATH:~/Documents/Datasift/Code/vedo-data-science-toolkit</pre>
+
+You execute the tool as follows:
 
     python scored_tags_classifier.py --test_period=[Test period] --config_module_path=[config file] --training_json=[training interactions] --training_csv=[OPTIONAL: label file] --classpath=[label path] > [output file]
 

linearclassification/lib/config_base.py

@@ -1,4 +1,4 @@
-from lib.features import *
+from linearclassification.lib.features import *
 
 class ConfigBase:
 

linearclassification/lib/default_config.py

@@ -1,5 +1,5 @@
-from lib.features import *
-from lib.config_base import ConfigBase
+from linearclassification.lib.features import *
+from linearclassification.lib.config_base import ConfigBase
 
 class Config(ConfigBase):
 

linearclassification/lib/features.py

@@ -1,5 +1,5 @@
 import re,sys
-from lib.utils import jpath,chunk,contains,has_subsequence
+from linearclassification.lib.utils import jpath,chunk,contains,has_subsequence
 
 wordsplitter=re.compile(r'[^\w\$]+')
 

linearclassification/scored_tags_classifier.py

@@ -5,11 +5,11 @@
 import numpy as np
 from sklearn.linear_model import SGDClassifier
 
-from lib.metrics import confusion_matrix
+from linearclassification.lib.metrics import confusion_matrix
 import itertools
 
-from lib.features import *
-from lib.utils import jpath,nvl,ngrams,all_zeroes,chunk
+from linearclassification.lib.features import *
+from linearclassification.lib.utils import jpath,nvl,ngrams,all_zeroes,chunk
 
 '''
 Run as follows
@@ -98,8 +98,8 @@ def report_confusion(interactions,targets,fvectors,title):
   confusion_matrix(expectedvsactuals)
   for i,(exp,act) in enumerate(expectedvsactuals):
     if exp!=act:
-      logging.info("exp:act (%s,%s): %s |%s",exp,act, nvl(jpath(featurepath,interactions[i])),\
-        '|'.join([selected_features[idx].string() for idx,satisfied in enumerate(fvectors[i]) if satisfied]))
+      logging.info("exp:act (%s,%s): %s |%s",exp,act, nvl(jpath(featurepath,interactions[i])).encode('utf-8','ignore'),\
+        '|'.join([selected_features[idx].string() for idx,satisfied in enumerate(fvectors[i]) if satisfied]).encode('utf-8','ignore'))
 
 def urldomain(url):
   domain=re.sub('https?://','',url)