def _bars(series, xticks, section_number, sub_section, display_name):
sub_section = util.header(section_number, sub_section,
"{} Bars Graph".format(display_name))
bars(series, sort_labels=True, xticks=xticks)
return sub_section
def get_features_importance(model, feature_names, top=10):
# graph
terminal.markdown_h2("Top {} Feature Importances".format(top))
print()
feature_importances = model.feature_importances_
df_feature_importances = pd.DataFrame({"importance": feature_importances}, index=feature_names) \
.sort_values('importance', ascending=False)
top_df = df_feature_importances.head(top)
graphs.bars(top_df["importance"],
percentage_on_top=False,
title='Feature Importances',
ylabel="Importance",
xlabel="Feature",
data_processed=True)
# break line
print("\n")
# table
terminal.markdown_h2("Feature Importances")
print()
table = [["Feature", "Importance"]]
content = []
for i, x in enumerate(feature_importances):
content.append([feature_names[i], round(x, 3)])
content.sort(key=lambda x: x[1], reverse=True)
table += content
terminal.markdown_table(table)
def _bars(series, section_number, sub_section, display_name):
sub_section = util.header(section_number, sub_section,
"{} Bars Graph".format(display_name))
dskc_graphs.bars(series,
title="{} Bars Graph".format(display_name),
xlabel=display_name)
return sub_section
def _top_words(words_series, top_words, section_number, sub_section,
display_name):
sub_section = util.header(
section_number, sub_section,
"{} Top {} Words".format(display_name, top_words))
graphs.bars(words_series,
title="Top {} words".format(top_words),
xlabel="Word",
percentage_on_top=True,
max_values=top_words)
return sub_section
def time_graphs(df,
column,
ylabel="",
year=True,
month=True,
day=True,
weekday=True):
'''
Display graphs by year, month, day and weekday
:param df: pandas dataframe
:param column: column name
:param ylabel:
:param year:
:param month:
:param day:
:param weekday:
:return:
'''
ylabel = ylabel.capitalize()
if year:
series = df[column + "_YEAR"]
title = "over the Years"
if ylabel:
title = ylabel + " " + title
title = title.capitalize()
bars(series,
sort_labels=True,
title=title,
xlabel="Year",
ylabel=ylabel)
if month:
series = df[column + "_MONTH"]
title = "over the Months"
if ylabel:
title = ylabel + " " + title
title = title.capitalize()
bars(series,
sort_labels=True,
title=title,
xlabel="Month",
ylabel=ylabel)
if day:
title = "over the Days"
if ylabel:
title = ylabel + " " + title
title = title.capitalize()
bars(df[column + "_DAY"],
title=title,
sort_labels=True,
xlabel="Day",
ylabel=ylabel)
if weekday:
series = df[column + "_WEEKDAY"]
title = "over the Weeks"
if ylabel:
title = ylabel + " " + title
title = title.capitalize()
days = [
'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday',
'Sunday'
]
bars(series,
sort_labels=True,
title=title,
xlabel="Day of the Week",
xticks=days,
ylabel=ylabel)