def search(new_grid, Cs):
grid = {
'segment': ['gender_female'],
'files_n': [5],
'lower_pop_bound': [3],
'ngram_ns': [[1, 2]],
'min_word_length': [2],
'word_type': ['alphanumeric'],
'use_urls': [True],
'min_url_depth': [1]
}
for k, v in new_grid.iteritems():
grid[k] = v
res = []
for p in params(grid):
X, y, X_test, y_test, features_map = \
constr(segment=p['segment'], files_n=p['files_n'],
lower_pop_bound=p['lower_pop_bound'],
ngram_ns=p['ngram_ns'], min_word_length=p['min_word_length'], word_type=p['word_type'],
use_urls=p['use_urls'], min_url_depth=p['min_url_depth'])
sc = set()
for C in Cs:
sc.add((C, C_score(C, X, y, X_test, y_test)))
res.append((p, X.shape, sc))
print res
return res
def search(new_grid, Cs):
grid = {'segment' : ['gender_female'], 'files_n' : [5],
'lower_pop_bound' : [3],
'ngram_ns' : [[1, 2]], 'min_word_length' : [2], 'word_type' : ['alphanumeric'],
'use_urls' : [True], 'min_url_depth' : [1]}
for k, v in new_grid.iteritems():
grid[k] = v
res = []
for p in params(grid):
X, y, X_test, y_test, features_map = \
constr(segment=p['segment'], files_n=p['files_n'],
lower_pop_bound=p['lower_pop_bound'],
ngram_ns=p['ngram_ns'], min_word_length=p['min_word_length'], word_type=p['word_type'],
use_urls=p['use_urls'], min_url_depth=p['min_url_depth'])
sc = set()
for C in Cs:
sc.add((C, C_score(C, X, y, X_test, y_test)))
res.append((p, X.shape, sc))
print res
return res
import cPickle as pickle
from constr import constr
X, y, X_test, y_test, features_map = constr(segment='gender_female', files_n=47,
lower_pop_bound=2,
ngram_ns=[1, 2], min_word_length=1, word_type='alphanumeric',
use_urls=True, min_url_depth=1)
pickle.dump((X, y, X_test, y_test, features_map), open('prepared/gender_female_allegedly_best.pickle', 'wb'), -1)