def graph(self, data, output_folder, parent_folder):
if self.type == None:
raise ValueError("Grapher type must be set to a string")
data = util.group_words_by_term(data, get_tfidf=True)
data = data[data['word'].str.len() > 1]
data = data[data['type'] == 'word']
terms = sorted(data.term.unique().tolist())
N = len(terms)
rows, cols = util.get_rows_cols(N)
fig, ax = plt.subplots(figsize=(cols * 2, rows * 3), ncols=cols, nrows=rows, squeeze=False)
plt.subplots_adjust(
left = 0.2,
bottom = 0.1,
right = 2,
top = 0.9,
wspace = 0.5,
hspace = 1.1
)
for i in range(N):
ax[int(i / cols)][i % cols].set_title(terms[i], y = 1)
to_plot = data[data['term'] == terms[i]]
to_plot = to_plot.head(10)[['term','word','tf_idf']]
plot = sns.barplot(
y=to_plot['word'],
x=to_plot['tf_idf'],
data=to_plot,
palette = config.PALETTE,
orient="h",
ax=ax[int(i / cols)][i % cols]
)
plot.set(
ylabel="",
xlabel="Distinctiveness Score"
)
TITLE = "Each Term's Most Distinguishing {}".format(self.type)
plt.suptitle(TITLE, y = 1.09, fontsize=20)
fig.savefig(
"{}/{}.png".format(output_folder, slugify(TITLE)),
bbox_inches='tight',
pad_inches=config.PAD_INCHES
)
fig.clf()
def graph(self, data, output_folder, parent_folder):
data = util.group_words_by_sender(data, get_tfidf=True)
data = data[data['word'].str.len() > 1]
data = data[data['type'] == 'word']
senders = data[config.SENDER_COLUMN_NAME].unique().tolist()
N = len(senders)
rows, cols = util.get_rows_cols(N)
fig, ax = plt.subplots(figsize=(cols * 2, rows * 3),
ncols=cols,
nrows=rows,
squeeze=False)
plt.subplots_adjust(left=0.2,
bottom=0.1,
right=2,
top=0.9,
wspace=0.5,
hspace=1.1)
for i in range(N):
ax[int(i / cols)][i % cols].set_title(senders[i], y=1)
to_plot = data[data[config.SENDER_COLUMN_NAME] == senders[i]]
to_plot = to_plot.head(10)[[
config.SENDER_COLUMN_NAME, 'word', 'tf_idf'
]]
plot = sns.barplot(y=to_plot['word'],
x=to_plot['tf_idf'],
data=to_plot,
palette=config.PALETTE,
orient="h",
ax=ax[int(i / cols)][i % cols])
plot.set(ylabel="", xlabel="Distinctiveness Score")
TITLE = "Our Most Distinguishing Words"
plt.suptitle(TITLE, y=1.09, fontsize=20)
fig.savefig("{}/{}.png".format(output_folder, slugify(TITLE)),
bbox_inches='tight',
pad_inches=config.PAD_INCHES)
fig.clf()