def plot_regression_predictions(y_test, model_name, prediction):
y_test1 = y_test.copy()
y_test1 = y_test1.reset_index()
# y_test1['Date'] = dt.strftime(y_test.index.get_level_values('Date'), "%Y-%m-%d")
y_test1['Date'] = dt.strftime(y_test.index.get_level_values('Date'),
"%Y-%m")
y_test1['Date'] = pd.to_datetime(y_test1['Date'])
y_test1.drop('POI', axis=1, inplace=True)
f, ax = plt.subplots(figsize=(8, 5))
ax.plot(y_test1.Date,
y_test1.Percentage_Bad_Words_count,
label='% of Total Email Messages')
ax.plot(y_test1.Date,
prediction,
color='red',
label="% of Financial Negative Words Predicted")
ax.plot_date(y_test1.Date, prediction, color='black')
ax.set_title(model_name +
': % of Financial Negative Words Predicted by Date')
ax.set_xlabel("Date", rotation=0)
plt.xticks(rotation=90)
ax.set_ylabel("% Of Negative Word")
ax.legend(loc="upper right")
f.autofmt_xdate()
plt.show()
def get_all_dates_to_download(
from_date: DatetimeIndex, to_date: DatetimeIndex,
time_resolution: TimeResolution) -> DatetimeIndex:
"""
Returns a DatetimeIndex with dates from_date to to_date in time_resolution.
Args:
from_date (DatetimeIndex): DatetimeIndex object setting the start of time interval.
to_date (DatetimeIndex): DatetimeIndex object setting the end of the time interval.
time_resolution (TimeResolution): Specifying the time resolution used in blob storage.
Returns:
Returns a pandas.DatetimeIndex with all dates from_date to to_date (both included) in time_resolution.
"""
# Get all the dates we need
if time_resolution == TimeResolution.NONE:
# We handle this case by making a DatetimeIndex containing one element. We will ignore the
# date anyway.
return pd.date_range(start='1/1/2021', periods=1, freq='D')
if time_resolution == TimeResolution.YEAR:
return pd.date_range(from_date.strftime("%Y"),
to_date.strftime("%Y"),
freq='AS')
if time_resolution == TimeResolution.MONTH:
return pd.date_range(from_date.strftime("%Y-%m"),
to_date.strftime("%Y-%m"),
freq='MS')
if time_resolution == TimeResolution.DAY:
return pd.date_range(from_date.strftime("%Y-%m-%d"),
to_date.strftime("%Y-%m-%d"),
freq='D')
if time_resolution == TimeResolution.HOUR:
return pd.date_range(from_date.strftime("%Y-%m-%dT%H"),
to_date.strftime("%Y-%m-%dT%H"),
freq='H')
if time_resolution == TimeResolution.MINUTE:
return pd.date_range(from_date.strftime("%Y-%m-%dT%H:%M"),
to_date.strftime("%Y-%m-%dT%H:%M"),
freq='T')
raise ValueError('(ValueError) Unknown time resolution given.')
def plot_exploration(email_data_frame_class):
plot_Data = email_data_frame_class.copy()
plot_Data['POI1'] = email_data_frame_class.index.get_level_values('POI')
plot_Data['Date1'] = dt.strftime(
email_data_frame_class.index.get_level_values('Date'), "%Y-%m-%d")
plot_Data['Year'] = DatetimeIndex(plot_Data['Date1']).year
plot_Data['Month'] = DatetimeIndex(plot_Data['Date1']).month
plot_Data['Day'] = DatetimeIndex(plot_Data['Date1']).day
plot_Data['Year'] = DatetimeIndex(plot_Data['Date1']).year
plot_Data['Year_Month'] = plot_Data["Year"].astype(
str) + '/' + plot_Data["Month"].astype(str)
plot_Data['Year_Month'] = sorted(
plot_Data['Year_Month'],
key=lambda x: datetime.datetime.strptime(x, '%Y/%m'))
# plot_Data['Year_Month'] = plot_Data["Year"].astype(str) + plot_Data["Month"].astype(str)
# plot_Data['Year_Month'] = plot_Data['Year_Month'].astype(int)
df3 = plot_Data.groupby(['Year']).count()
df4 = plot_Data.groupby(['Year_Month']).count()
df5 = plot_Data[plot_Data['Bad_Words_count'] > 0]
df5 = df5.groupby(['Year_Month']).count()
ax = df3.file.plot(kind='bar',
title="Message By Year",
figsize=(5, 5),
fontsize=8)
ax.set_xlabel("Year", fontsize=8)
ax.set_ylabel("Number Of Message", fontsize=8)
plt.xticks(rotation=90)
plt.grid()
plt.show()
f, ax = plt.subplots(figsize=(15, 10))
ax = df4.file.plot(kind='bar',
title="Count Of Messages By Year Month",
fontsize=10)
df4.file.plot(kind='line',
ax=ax,
color='green',
label='Total Messages',
linewidth=3)
ax.legend(loc='upper center', shadow=True)
plt.xticks(rotation=90)
ax.set_xticklabels(
sorted(df5.index.get_level_values('Year_Month'),
key=lambda x: datetime.datetime.strptime(x, '%Y/%m')))
ax.set_ylabel("Number Of Message", fontsize=8)
plt.grid()
plt.show()
f, ax = plt.subplots(figsize=(15, 10))
ax = df5.file.plot(
kind='bar',
title="Count Of Messages With Negative Word By Year Month",
fontsize=10)
df5.file.plot(kind='line',
ax=ax,
color='red',
label='Messages With Negative Words',
linewidth=5)
ax.legend(loc='upper center', shadow=True)
plt.xticks(rotation=90)
ax.set_xticklabels(
sorted(df5.index.get_level_values('Year_Month'),
key=lambda x: datetime.datetime.strptime(x, '%Y/%m')))
ax.set_ylabel("Number Of Message With Negative Words", fontsize=8)
plt.grid()
plt.show()
