def partition_info_table(viomet_df, date_range, partition_infos):
'''
TODO
'''
index_keys = [('MSNBC', 'MSNBCW'), ('CNN', 'CNNW'),
('Fox News', 'FOXNEWSW')]
columns = [
'$t_0^{(2)}$', '$t^{(2)}_{N^{(2)}}$', '$f^{(1)}$', '$f^{(2)}$',
'reactivity', 'total uses'
]
counts_df = daily_metaphor_counts(viomet_df, date_range, by=['network'])
data = []
for ik in index_keys:
key = ik[1]
pi = partition_infos[key]
data.append([
pi.partition_date_1, pi.partition_date_2,
pi.f_ground, pi.f_excited,
((pi.f_excited - pi.f_ground) / pi.f_ground), counts_df[key].sum()
])
index = [ik[0] for ik in index_keys]
return pd.DataFrame(data=data, index=index, columns=columns)
def by_network_word_table(project_df,
date_range,
partition_infos,
words=['hit', 'beat', 'attack']):
'''
Second table in paper
'''
networks = ['MSNBC', 'CNN', 'Fox News']
columns = ['fg', 'fe', 'reactivity', 'total uses']
# index_tuples = [(net, word) for net in networks for word in words]
index_tuples = [(word, net) for word in words for net in networks]
index = pd.MultiIndex.from_tuples(index_tuples,
names=['Violent Word', 'Network'])
df = pd.DataFrame(index=index, columns=columns, data=0.0)
project_df = project_df[project_df.include]
counts_df = daily_metaphor_counts(project_df,
date_range,
by=['network', 'facet_word'])
for idx, netid in enumerate(['MSNBCW', 'CNNW', 'FOXNEWSW']):
sum_g, n_g = _get_ground(counts_df,
netid,
partition_infos,
words=words)
sum_e, n_e = _get_excited(counts_df,
netid,
partition_infos,
words=words)
freq_g = sum_g / n_g
freq_e = sum_e / n_e
reactivity = ((freq_e - freq_g) / freq_g)
totals = sum_g + sum_e
network = networks[idx]
for word in words:
df.loc[word, network] = [
freq_g[word], freq_e[word], reactivity[word], totals[word]
]
fancy_columns = ['$f^{(1)}$', '$f^{(2)}$', 'reactivity', 'total uses']
df.columns = fancy_columns
return df
def by_network_subj_obj_table(viomet_df,
date_range,
partition_infos,
subjects=['Barack Obama', 'Mitt Romney'],
objects=['Barack Obama', 'Mitt Romney']):
'''
TODO
'''
networks = ['MSNBC', 'CNN', 'Fox News']
columns = ['fg', 'fe', 'reactivity', 'total uses']
# index_tuples = [(net, word) for net in networks for word in words]
subj_objs = ["Subject=" + subj for subj in subjects] \
+ ["Object=" + obj for obj in objects]
index_tuples = [(so, net) for so in subj_objs for net in networks]
index = pd.MultiIndex.from_tuples(index_tuples,
names=['Subject/Object', 'Network'])
df = pd.DataFrame(index=index, columns=columns, data=0.0)
# Next two blocks support more than two subjects or objects.
subject_rows = reduce(
lambda x, y: (viomet_df.subjects == x) | (viomet_df.subjects == y),
subjects)
object_rows = reduce(
lambda x, y: (viomet_df.objects == x) | (viomet_df.objects == y),
objects)
subject_df = viomet_df[subject_rows]
object_df = viomet_df[object_rows]
subject_counts_df = daily_metaphor_counts(
subject_df,
date_range,
by=['network', 'subjects'],
)
object_counts_df = daily_metaphor_counts(object_df,
date_range,
by=['network', 'objects'])
for idx, network_id in enumerate(['MSNBCW', 'CNNW', 'FOXNEWSW']):
# Ground state data.
sum_subj_g, n_subj_g = _get_ground(subject_counts_df, network_id,
partition_infos)
sum_obj_g, n_obj_g = _get_ground(object_counts_df, network_id,
partition_infos)
# Excited state data.
sum_subj_e, n_subj_e = _get_excited(subject_counts_df, network_id,
partition_infos)
sum_obj_e, n_obj_e = _get_excited(object_counts_df, network_id,
partition_infos)
freq_subj_g = sum_subj_g / n_subj_g
freq_obj_g = sum_obj_g / n_obj_g
freq_subj_e = sum_subj_e / n_subj_e
freq_obj_e = sum_obj_e / n_obj_e
reactivity_diff_subj = ((freq_subj_e - freq_subj_g) / 2.0)
reactivity_diff_obj = ((freq_obj_e - freq_obj_g) / 2.0)
totals_subj = sum_subj_g + sum_subj_e
totals_obj = sum_obj_g + sum_obj_e
network = networks[idx]
for subject in subjects:
df.loc["Subject=" + subject, network] = [
freq_subj_g[subject], freq_subj_e[subject],
reactivity_diff_subj[subject], totals_subj[subject]
]
for object_ in objects:
df.loc["Object=" + object_, network] = [
freq_obj_g[object_], freq_obj_e[object_],
reactivity_diff_obj[object_], totals_obj[object_]
]
fancy_columns = ['$f^{(1)}$', '$f^{(2)}$', 'reactivity', 'total uses']
df.columns = fancy_columns
return df