In [6]:

    

import IPython
import numpy as np
import pandas as pd
from tsa.science import numpy_ext as npx

import viz
import viz.stats
import viz.format

from sklearn import metrics, cross_validation
from sklearn import linear_model
from sklearn import naive_bayes
from sklearn import svm
from sklearn import feature_selection

from tsa import stdout, stderr
from tsa.lib import tabular, datetime_extra
from tsa.lib.timer import Timer
from tsa.models import Source, Document, create_session
from tsa.science import features, models, timeseries
from tsa.science.corpora import MulticlassCorpus
from tsa.science.plot import plt, figure_path, distinct_styles, ticker
from tsa.science.summarization import metrics_dict, metrics_summary, average_accuracy


    

In [2]:

    

documents = Source.from_name('sb5b')
full_corpus = MulticlassCorpus(documents)
full_corpus.apply_labelfunc(lambda doc: doc.label)


    

In [4]:

    

full_corpus.labels


    

    
Out[4]:





array([u'For', u'Broken Link', u'Against', u'Neutral', None,
       u'Not Applicable'], dtype=object)

In [15]:

    

tri_classes = [full_corpus.class_lookup[label] for label in ['For', 'Against', 'Neutral']]
tri_indices = np.in1d(full_corpus.y, tri_classes)
tri_corpus = full_corpus.subset(tri_indices)
tri_corpus.extract_features(lambda doc: 1, features.intercept)
tri_corpus.extract_features(lambda doc: doc.document, features.ngrams,
    ngram_max=2, min_df=2, max_df=1.0)


    

    
Out[15]:





array([    1,     2,     3, ..., 43694, 43695, 43696])

In [16]:

    

npx.table(tri_corpus.y)


    

    
Out[16]:





[(0, 2785), (1, 0), (2, 10842), (3, 149)]

In [7]:

    

average_accuracy(tri_corpus, linear_model.LogisticRegression(fit_intercept=False, penalty='l2'))


    

    
Out[7]:





0.9496736765772299

In [27]:

    

folds = list(cross_validation.StratifiedShuffleSplit(tri_corpus.y, test_size=0.75, n_iter=10))

corpus = tri_corpus
train_indices, test_indices = folds[0]
train_corpus = corpus.subset(train_indices)
test_corpus = corpus.subset(test_indices)
        
model = linear_model.LogisticRegression(fit_intercept=False, penalty='l1')
model.fit(train_corpus.X, train_corpus.y)
pred_y = model.predict(test_corpus.X)

confusion_counts = metrics.confusion_matrix(test_corpus.y, pred_y)
print 'overall accuracy', metrics.accuracy_score(test_corpus.y, pred_y)
print 'true labels on the rows, predicted labels in the columns'
pd.DataFrame(confusion_counts, index=corpus.labels[tri_classes], columns=corpus.labels[tri_classes])


    

    




overall accuracy 0.92809445466






    
Out[27]:






  

    

      

      
For
      
Against
      
Neutral
    
  
  

    

      
For
      
 1604
      
  485
      
 0
    
    

      
Against
      
  148
      
 7984
      
 0
    
    

      
Neutral
      
   14
      
   96
      
 2
    
  

3 rows × 3 columns

In [23]:

    




    

    
Out[23]:





array([u'For', u'Against', u'Neutral'], dtype=object)

In [22]:

    

tri_classes


    

    
Out[22]:





[0, 2, 3]

In [ ]: