def top_entry_bar_plot(self, save=False):
"""
Bar chart of turnstile entrance data for top stations.
:param bool save: if True the figure will be saved
"""
fig = plt.figure('Top Station Box Plot',
figsize=(8, 5),
facecolor='white',
edgecolor='black')
rows, cols = (1, 1)
ax0 = plt.subplot2grid((rows, cols), (0, 0))
(self.top_stations.sort_values(by='entry').loc[:,
'Monday':'Sunday'].plot(
kind='bar',
alpha=0.5,
edgecolor='black',
ax=ax0))
ax0.set_title(f'{self.date_start} - {self.date_end}',
fontsize=size['title'])
ax0.set_xlabel('Subway Station', fontsize=size['label'])
ax0.set_xticklabels(ax0.xaxis.get_majorticklabels(), rotation=80)
ax0.set_ylabel('Turnstile Entries ($billions$)',
fontsize=size['label'])
ax0.yaxis.set_major_formatter(ax_formatter['billions'])
plt.tight_layout()
plt.suptitle('Turnstile Data',
fontsize=size['super_title'],
x=0.54,
y=1.05)
save_fig('age_voted', save)
def train_plot(self, save=False):
"""
Plot subway stations by train.
:param bool save: if True the figure will be saved
"""
sns.lmplot(x='longitude',
y='latitude',
data=(self.trains.sort_values(by='train')),
hue='train',
fit_reg=False,
legend=False,
markers='d',
scatter_kws={'alpha': 0.3},
size=10)
legend = plt.legend(bbox_to_anchor=(1.10, 0.5),
fancybox=True,
fontsize=size['legend'],
loc='center right',
shadow=True,
title='Train')
plt.setp(legend.get_title(), fontsize=size['label'])
plt.xlabel('Longitude', fontsize=size['label'])
plt.ylabel('Latitude', fontsize=size['label'])
plt.tight_layout()
plt.suptitle('New York Subway Train Stations',
fontsize=size['super_title'],
y=1.03)
save_fig('stations_trains', save)
def targets_entry_plot(self, save=False):
"""
Bar chart of turnstile entrance data.
:param bool save: if True the figure will be saved
"""
fig = plt.figure('Station Box Plot',
figsize=(8, 10),
facecolor='white',
edgecolor='black')
rows, cols = (2, 1)
ax0 = plt.subplot2grid((rows, cols), (0, 0))
ax1 = plt.subplot2grid((rows, cols), (1, 0))
# Pie Plot
pie = (self.target_data.groupby('station')['entry'].sum().sort_values(
ascending=False))
(pie.plot(kind='pie',
colors=sns.color_palette('gnuplot', n_colors=10),
explode=[0.05] * pie.shape[0],
labels=None,
legend=None,
shadow=True,
startangle=180,
ax=ax0))
ax0.set_aspect('equal')
ax0.set_ylabel('')
ax0.legend(bbox_to_anchor=(1.3, 0.5), labels=pie.index, loc='center')
# Bar Plot
(self.target_data.groupby('station')[[
'entry', 'exit'
]].sum().sort_values(by='entry').plot(kind='bar',
alpha=0.5,
edgecolor='black',
ax=ax1))
ax1.set_title(f'{self.date_start} - {self.date_end}',
fontsize=size['title'])
ax1.legend(['Entry', 'Exit'])
ax1.set_xlabel('Subway Station', fontsize=size['label'])
ax1.set_xticklabels(ax0.xaxis.get_majorticklabels(), rotation=80)
ax1.set_ylabel('Turnstile Use ($billions$)', fontsize=size['label'])
ax1.yaxis.set_major_formatter(ax_formatter['billions'])
plt.tight_layout()
plt.suptitle('Turnstile Data',
fontsize=size['super_title'],
x=0.53,
y=1.05)
save_fig('age_voted', save)
def daily_use_plot(self, station_name, save=False):
"""
Plot average daily turnstile use.
:param str station_name: station name
:param bool save: if True the figure will be saved
"""
if self.target_data is not station_name:
self.average_daily_use(station_name)
fig = plt.figure('Station By Day',
figsize=(12, 12),
facecolor='white',
edgecolor='black')
rows, cols = (3, 1)
ax0 = plt.subplot2grid((rows, cols), (0, 0))
ax1 = plt.subplot2grid((rows, cols), (1, 0), sharex=ax0, sharey=ax0)
ax2 = plt.subplot2grid((rows, cols), (2, 0), sharex=ax0, sharey=ax0)
bar_plot = (self.daily_use.loc[(
slice(0, 7),
slice(
pd.to_datetime('09:00:00').time(),
pd.to_datetime('17:00:00').time())), :].reset_index().rename(
columns={
'date_time': 'weekday',
'level_1': 'time'
}))
bar_plot['day_name'] = (
bar_plot.weekday.map(lambda x: f'{self.days[x]}'))
bar_plot['entry_legend'] = (
bar_plot.weekday.map(lambda x: f'{self.days[x]} Entry'))
bar_plot['exit_legend'] = (
bar_plot.weekday.map(lambda x: f'{self.days[x]} Exit'))
# entry data
sns.barplot(x='time',
y='entry',
hue='day_name',
data=bar_plot,
alpha=0.7,
edgecolor='black',
palette='gnuplot',
ax=ax0)
ax0.set_title('Turnstile Entrances', fontsize=size['title'])
# exit data
sns.barplot(x='time',
y='exit',
hue='day_name',
data=bar_plot,
alpha=0.5,
edgecolor='black',
ax=ax1)
ax1.set_title('Turnstile Exits', fontsize=size['title'])
# combined plot
sns.barplot(x='time',
y='entry',
hue='entry_legend',
data=bar_plot,
alpha=0.7,
edgecolor='black',
palette='gnuplot',
ax=ax2)
sns.barplot(x='time',
y='exit',
hue='exit_legend',
data=bar_plot,
alpha=0.5,
edgecolor='black',
ax=ax2)
ax2.set_title('Turnstile Entrance and Exits', fontsize=size['title'])
for ax in (ax0, ax1, ax2):
ax.legend(bbox_to_anchor=(1.2, 1),
fontsize=size['legend'],
loc='upper right')
ax.set_xlabel('Time of Day', fontsize=size['label'])
ax.set_ylabel('Turnstile Use ($millions$)', fontsize=size['label'])
ax.yaxis.set_major_formatter(ax_formatter['millions'])
plt.tight_layout()
plt.suptitle(
f'Station Average Turnstile Use: '
f'{station_name.title()}\n'
f'{self.date_start} - {self.date_end}\n',
fontsize=size['super_title'],
x=0.52,
y=1.07)
save_fig('daily_use_plot', save)
def age_vote_plot(self, save=False):
"""
US Voter Age plots
:param bool save: if True the figure will be saved
"""
fig = plt.figure('Voter Age',
figsize=(12, 10),
facecolor='white',
edgecolor='black')
rows, cols = (2, 2)
ax0 = plt.subplot2grid((rows, cols), (0, 0))
ax1 = plt.subplot2grid((rows, cols), (0, 1), sharex=ax0, sharey=ax0)
ax2 = plt.subplot2grid((rows, cols), (1, 0), sharex=ax0)
ax3 = plt.subplot2grid((rows, cols), (1, 1), sharex=ax0, sharey=ax2)
both_sexes = self.data.query('sex == "both sexes"')
female = self.data.query('sex == "female"')
male = self.data.query('sex == "male"')
# Voters All US Plot
(both_sexes.voted_yes.plot(kind='area',
alpha=0.5,
label='Voters',
ax=ax0))
target_group = both_sexes.voted_yes.copy()
target_group.loc[target_group < target_group.quantile(0.7)] = 0
(target_group.plot(kind='area',
alpha=0.2,
color='C2',
label='Target Group',
ax=ax0))
ax0.set_title('Voters vs Age', fontsize=size['title'])
ax0.legend(loc='upper left', fontsize=size['legend'])
ax0.set_ylabel('People ($thousands$)', fontsize=size['label'])
ax0.yaxis.set_major_formatter(ax_formatter['thousands'])
# Voters Gender Plot
for n, gender in enumerate((female, male)):
(gender.voted_yes.plot(kind='area',
alpha=0.5,
color=f'C{n}',
ax=ax1))
ax1.set_title('Voters vs Age by Gender', fontsize=size['title'])
# Voters All US Percentage Plot
(both_sexes.voted_yes.div(both_sexes.us_population).plot(
kind='area', alpha=0.5, label='Voter Percentage', ax=ax2))
target_group = both_sexes.voted_yes_pct.copy()
target_group.loc[target_group < 0.5] = 0
(target_group.plot(kind='area',
alpha=0.2,
color='C2',
label='Target Group',
ax=ax2))
ax2.set_title('Percent Voters vs Age', fontsize=size['title'])
ax2.legend(loc='upper left', fontsize=size['legend'])
ax2.set_ylabel('Voting Percentage of Age', fontsize=size['label'])
ax2.yaxis.set_major_formatter(ax_formatter['percent_convert'])
# Voters Gender Percentage Plot
for n, gender in enumerate((female, male)):
(gender.voted_yes.div(gender.us_population).plot(kind='area',
alpha=0.5,
color=f'C{n}',
ax=ax3))
ax3.set_title('Percent Voters vs Age by Gender',
fontsize=size['title'])
for ax in (ax0, ax1):
ax.set_xlabel('')
for ax in (ax2, ax3):
ax.set_xlabel('Age ($years$)', fontsize=size['label'])
for ax in (ax1, ax3):
ax.legend(['Female', 'Male'], fontsize=size['legend'])
plt.tight_layout()
plt.suptitle('2016 US Voter Age Distributions',
fontsize=size['super_title'],
y=1.03)
save_fig('age_voted', save)
def ethnicity_plot(self, save=False):
"""
New York state ethnicity voter plots.
:param bool save: if True the figure will be saved
"""
fig = plt.figure('Age Vote',
figsize=(10, 10),
facecolor='white',
edgecolor='black')
rows, cols = (2, 1)
ax0 = plt.subplot2grid((rows, cols), (0, 0))
ax1 = plt.subplot2grid((rows, cols), (1, 0))
# Voter Turnout
(self.data.loc[
'new york',
['race', 'state_population', 'registered_yes', 'voted_yes']].
set_index('race').sort_values(by='voted_yes').plot(kind='bar',
alpha=0.5,
edgecolor='black',
ax=ax0))
ax0.set_title('2016 New York Voters', fontsize=size['title'])
ax0.legend(['Total Population', 'Registered Voters', 'Voted in 2016'],
fontsize=size['legend'])
ax0.set_ylabel('People ($thousands$)', fontsize=size['label'])
ax0.yaxis.set_major_formatter(ax_formatter['thousands'])
# Voter Turnout Percentage
voter_pct = (self.data.loc['new york', [
'race', 'registered_total_pct', 'registered_total_error',
'voted_total_pct', 'voted_total_error'
]].set_index('race').applymap(lambda x: float(x)).sort_values(
by='voted_total_pct'))
data_cols = ['registered_total_pct', 'voted_total_pct']
voter_err = voter_pct.drop(data_cols, axis=1)
voter_err.columns = data_cols
error_kw = {'capsize': 3, 'capthick': 1, 'elinewidth': 1}
voter_pct.plot(kind='bar',
y=data_cols,
alpha=0.5,
edgecolor='black',
error_kw=error_kw,
yerr=voter_err,
ax=ax1)
ax1.set_title('2016 New York Voter Percentages',
fontsize=size['title'])
ax1.legend(['Registered Voters', 'Voted in 2016'],
fontsize=size['legend'])
ax1.set_ylabel('Percentage', fontsize=size['label'])
ax1.yaxis.set_major_formatter(ax_formatter['percent'])
for ax in (ax0, ax1):
ax.set_xlabel('Group', fontsize=size['label'])
ax.set_xticklabels(ax.xaxis.get_majorticklabels(),
fontsize=12,
rotation=75)
plt.tight_layout()
plt.suptitle('2016 New York State Voter Distributions',
fontsize=size['super_title'],
y=1.03)
save_fig('age_voted', save)