def display_handler(self, *_):
if len(self.data) == 0:
self.alert("No data!")
return
self._output.clear_output()
with self._output:
if self.output_format.lower() == 'chart':
top_tokens: pd.DataFrame = self.data.assign(
xs=self.data.index, ys=self.data.weight)
p = plot_topic_word_distribution(
top_tokens,
plot_width=1200,
plot_height=500,
title='',
tools='box_zoom,wheel_zoom,pan,reset')
bokeh.plotting.show(p)
else:
if self.output_format.lower() in ('xlsx', 'csv', 'clipboard'):
utility.ts_store(data=self.data,
extension=self.output_format.lower(),
basename='topic_word_distribution')
g: gu.TableWidget = gu.table_widget(self.data)
display(g)
self.alert("✅")
def default_displayer(opts: "TopicsTokenNetworkGUI") -> None:
if opts.model.top_topic_tokens is None:
return
topics_tokens = opts.model.get_topics_tokens(opts.topics_ids,
opts.top_count)
if opts.output_format == "network":
network = create_network(topics_tokens)
opts.network = network
opts.set_layout()
css_style = css_styles(topics_tokens.topic_id.unique(),
opts.custom_styles)
network.set_style(css_style)
display(network)
if opts.output_format.lower() in ('table', 'xlsx', 'csv', 'clipboard',
'gephi'):
if opts.output_format == "gephi":
topics_tokens = topics_tokens[['topic', 'token', 'weight']]
topics_tokens.columns = ['Source', 'Target', 'Weight']
if opts.output_format != "table":
utility.ts_store(data=topics_tokens,
extension=opts.output_format.lower(),
basename='topics_token_network')
g = gu.table_widget(topics_tokens)
display(g)
def update_handler(self, *_):
self.alert("⌛ Computing...")
self._output.clear_output()
try:
with self._output:
try:
weights: pd.DataFrame = self.update()
except pu.EmptyDataError:
weights = None
if weights is None:
self.alert("😡 No data, please change filters..")
elif self.output_format in ('xlsx', 'csv', 'clipboard'):
pu.ts_store(data=weights,
extension=self.output_format,
basename='heatmap_weights')
elif self.output_format == "table":
g = gu.table_widget(weights)
display(g)
else:
display_heatmap(
weights,
self.titles,
flip_axis=self._flip_axis.value,
aggregate=self.aggregate,
output_format=self.output_format,
)
self.alert("✅")
except Exception as ex:
self.warn(f"😡 {ex}")
def display_wordcloud(
inferred_topics: tm.InferredTopicsData,
topic_id: int = 0,
n_words: int = 100,
output_format: str = 'Wordcloud',
):
tokens = inferred_topics.get_topic_title(topic_id, n_tokens=n_words)
if len(tokens) == 0:
raise ValueError("No data! Please change selection.")
if output_format == 'Wordcloud':
plot_utility.plot_wordcloud(inferred_topics.get_topic_tokens(topic_id),
token='token',
weight='weight',
max_words=n_words,
**PLOT_OPTS)
else:
top_tokens: pd.DataFrame = inferred_topics.get_topic_top_tokens(
topic_id=topic_id, n_tokens=n_words)
g = gu.table_widget(top_tokens)
display(g)
if output_format.lower() in ('xlsx', 'csv', 'clipboard'):
utility.ts_store(data=top_tokens,
extension=output_format.lower(),
basename='topic_top_tokens')
def display_heatmap(
weights: pd.DataFrame,
titles: pd.DataFrame,
key: str = 'max', # pylint: disable=unused-argument
flip_axis: bool = False,
glyph: str = 'Circle', # pylint: disable=unused-argument
aggregate: str = None,
output_format: str = None,
):
''' Display aggregate value grouped by year '''
try:
weights['weight'] = weights[aggregate]
weights['year'] = weights.year.astype(str)
weights['topic_id'] = weights.topic_id.astype(str)
if len(weights) == 0:
print("No data! Please change selection.")
return
if output_format.lower() == 'heatmap':
p = plot_topic_relevance_by_year(
weights,
xs='year',
ys='topic_id',
flip_axis=flip_axis,
titles=titles,
element_id='topic_relevance',
**HEATMAP_FIGOPTS,
)
bokeh.plotting.show(p)
elif output_format.lower() in ('xlsx', 'csv', 'clipboard'):
pu.ts_store(data=weights,
extension=output_format.lower(),
basename='heatmap_weights')
else:
g = gu.table_widget(weights)
display(g)
except Exception as ex:
# raise
logger.error(ex)
def display_handler(self, *_):
self._output.clear_output()
try:
with self._output:
if self.yearly_topic_weights is None:
self.alert("😡 No data, please change filters..")
elif self.output_format in ('xlsx', 'csv', 'clipboard'):
pu.ts_store(data=self.yearly_topic_weights,
extension=self.output_format,
basename='heatmap_weights')
elif self.output_format == "pandas":
with pd.option_context("display.max_rows", None,
"display.max_columns", None):
display(self.yearly_topic_weights)
elif self.output_format == "table":
g = gu.table_widget(self.yearly_topic_weights,
handler=self.click_handler)
display(g)
else:
self.topic_changed()
self.alert("✅")
except Exception as ex:
self.warn(f"😡 {ex}")
def display_handler(self, *_):
self._output.clear_output()
with self._output:
if self.network_data is None:
self.alert("😡 No data, please change filters..")
elif self.output_format in ('xlsx', 'csv', 'clipboard', 'table',
'gephi'):
data: pd.DataFrame = self.network_data
if self.output_format == "gephi":
data = data[['topic_id', "title", 'weight']]
data.columns = ['Source', 'Target', 'Weight']
if self.output_format != "table":
pu.ts_store(data=data,
extension=self.output_format,
basename='heatmap_weights')
g: gu.TableWidget = gu.table_widget(data)
display(g)
else:
nx.plot_highlighted_bipartite_dataframe(
network_data=self.network_data,
network_layout=self.network_layout,
highlight_topic_ids=self.highlight_ids,
titles=self.titles,
scale=self.scale,
source_name="title", # FIXME:self.picked_pivot_name
target_name="topic_id",
element_id=self.text_id,
)
def display_handler(self, *_):
self._output.clear_output()
with self._output:
if self.network_data is None:
self.alert("😡 No data, please change filters..")
elif self.output_format in ('xlsx', 'csv', 'clipboard', 'table', 'gephi'):
data: pd.DataFrame = self.network_data
if self.output_format == "gephi":
data = data[['topic_id', "title", 'weight']]
data.columns = ['Source', 'Target', 'Weight']
if self.output_format != "table":
pu.ts_store(data=data, extension=self.output_format, basename='heatmap_weights')
g: gu.TableWidget = gu.table_widget(data)
display(g)
else:
display_topic_topic_network(
data=self.network_data,
layout=self._network_layout.value,
titles=self.titles,
scale=self._scale.value,
element_id=self.text_id,
node_range=self._node_range.value,
edge_range=self._edge_range.value,
topic_proportions=self.topic_proportions,
)
