def test_empty_breaks():
x = []
assert custom_format()(x) == []
assert comma_format()(x) == []
assert currency_format()(x) == []
assert percent_format()(x) == []
assert scientific_format()(x) == []
assert date_format()(x) == []
assert mpl_format()(x) == []
assert log_format()(x) == []
assert timedelta_format()(x) == []
def test_empty_breaks():
x = []
assert custom_format()(x) == []
assert comma_format()(x) == []
assert currency_format()(x) == []
assert percent_format()(x) == []
assert scientific_format()(x) == []
assert date_format()(x) == []
assert mpl_format()(x) == []
assert log_format()(x) == []
assert timedelta_format()(x) == []
def test_percent_format():
formatter = percent_format()
# same/nearly same precision values
assert formatter([.12, .23, .34, .45]) == \
['12%', '23%', '34%', '45%']
assert formatter([.12, .23, .34, 4.5]) == \
['12%', '23%', '34%', '450%']
# mixed precision values
assert formatter([.12, .23, .34, 45]) == \
['10%', '20%', '30%', '4500%']
def test_percent_format():
formatter = percent_format()
# same/nearly same precision values
assert formatter([.12, .23, .34, .45]) == \
['12%', '23%', '34%', '45%']
assert formatter([.12, .23, .34, 4.5]) == \
['12%', '23%', '34%', '450%']
# mixed precision values
assert formatter([.12, .23, .34, 45]) == \
['10%', '20%', '30%', '4500%']
def plot_elos():
diffs = np.linspace(-1000, +1000)
rates = 1 / (1 + 10**(-diffs / 400))
df = pd.DataFrame({'elo': diffs, 'winrate': rates})
return (pn.ggplot(df) + pn.geom_line(pn.aes(x='elo', y='winrate')) +
pn.geom_vline(xintercept=0, alpha=.1) +
pn.geom_hline(yintercept=.5, alpha=.1) +
pn.labs(x='Own Elo relative to opponent\'s Elo',
y='Win rate v. opponent') +
pn.scale_y_continuous(labels=percent_format()) +
pn.coord_cartesian(expand=False) + plot.IEEE())
def plot_calibrations():
params = data.sample_calibrations()
return (
pn.ggplot(
params,
pn.aes(xmin='boardsize-.25',
xmax='boardsize+.25',
group='boardsize',
fill='factor(boardsize)')) +
pn.geom_hline(yintercept=.5, alpha=.2) + pn.geom_rect(
pn.aes(ymin='lower', ymax='upper'), show_legend=False, color='k') +
pn.geom_rect(pn.aes(ymin='mid', ymax='mid'),
show_legend=False,
color='k',
size=2) + pn.scale_y_continuous(labels=percent_format()) +
pn.scale_fill_hue(l=.4) + pn.coord_cartesian(ylim=(.4, .6)) +
pn.labs(y='Win rate v. perfect play', x='Board size') + plot.IEEE())
def plot_importance_lgb(importance):
# Ugly but pip install problem on Airflow otherwise
import plotnine as pn
from plotnine import ggplot, aes # noqa
# from plotnine.geoms import * # noqa
coef = 1.5
pn.options.figure_size = (6.4 * coef, 4.8 * coef)
from mizani.formatters import percent_format # noqa
# from mizani.breaks import date_breaks # noqa
# from mizani.formatters import date_format # noqa
importance['importance'] = importance['importance'] / 100
importance['feature'] = pd.Categorical(importance['feature'],
importance['feature'][::-1],
ordered=True)
plot = (ggplot(importance, aes('feature', 'importance')) +
pn.geom_bar(stat='identity') + pn.coords.coord_flip() +
pn.scales.scale_y_continuous(labels=percent_format()) +
pn.labs(title='Feature importance', x='Feature', y='Gain'))
return plot
xax = axes[-1].xaxis
xax.set_ticklabels(lf(10.0**ex) for ex in xax.get_ticklocs())
plt.xlabel("growth rate [1/h]")
plt.ylabel("tradeoff")
plt.savefig("figures/dists.svg")
plt.close()
non_zero = (rates.groupby([
"id", "tradeoff"
]).apply(lambda df: (df.growth_rate > 1e-6).sum() / df.shape[0]).reset_index(
name="non_zero"))
pl = (ggplot(non_zero, aes(x="tradeoff", y="non_zero")) +
geom_boxplot(outlier_color="none") +
geom_jitter(width=0.15, height=0, alpha=0.5, stroke=0) +
scale_y_continuous(labels=percent_format()) +
labs(x="tradeoff", y="percent taxa growing"))
pl.save("figures/percent_growing.svg", width=5, height=5)
# Show some representative correlations
comp = both[(both.tradeoff == "0.5")]
w = within_samples[within_samples.tradeoff == "0.5"]
w.index = w.id
comp = comp[comp.id.isin(w.id[(w.n >= 10)])]
comp["rho"] = w.loc[comp.id, "rho"].round(2).values
comp.id = comp.id + " (r=" + comp.rho.astype(str) + ")"
pl = (ggplot(comp, aes(x="rate", y="growth_rate")) + geom_point() +
facet_wrap("~ id", nrow=3) + scale_x_log10() + scale_y_log10() +
labs(x="replication rate [a.u.]", y="predicted growth rate [1/h]"))
pl.save("figures/corr_examples.png", width=12, height=6, dpi=300)
gg_rep_act.save(os.path.join(dir_output, 'gg_rep_act.png'), width=8, height=4)
di_notes = {
'chi2': 'χ2-correction',
'insig': 'Erroneous',
'specification': 'Specification',
'non-replicable': 'Inconsistent'
}
# (ii) Breakdown of counts
tmp = acc_tt.merge(
res_fisher.tt.value_counts().reset_index().rename(columns={
'index': 'tt',
'tt': 'n_lit'
}))
tmp = tmp.assign(tt=lambda x: x.tt.map(di_tt),
notes=lambda x: x.notes.map(di_notes),
share=lambda x: x.n / x.n_lit)
gg_acc_notes = (
pn.ggplot(tmp, pn.aes(x='notes', y='share', fill='tt')) + pn.theme_bw() +
pn.scale_y_continuous(labels=percent_format(), limits=[0, 0.1]) +
pn.scale_fill_discrete(name='Literature') +
pn.geom_col(color='black', position=pn.position_dodge(0.5), width=0.5) +
pn.labs(y='Percent', x='Investigation') +
pn.theme(axis_text_x=pn.element_text(angle=45),
axis_title_x=pn.element_blank()))
gg_acc_notes.save(os.path.join(dir_output, 'gg_acc_notes.png'),
width=7,
height=3)
print('~~~ End of 4_results_insig.py ~~~')
def density_plot(df_original, variable, target=False, no_outliers=False, title=None, mean=True, mode=True, q1=True, q3=True):
'''
:param df: dataframe, variable: column to plot, target: object variable for split graph
:return: ggplot graphs of univariate analysis
by dtype: histogram, frequency, unique categories,
:usage: after cleaning dataframe
:target: is a string with the name of the categorical variable selected as target variable
'''
# target as str
df = df_original.copy()
df[target] = df[target].apply(str)
x_label = variable.lower().replace("_", " ").title()
graph = (ggplot(df) + aes(x=variable, y='..scaled..', fill=target))
# target
if target == False:
graph += geom_density(fill=colors.FIRST_COLOR)
else:
fill_label = target.lower().replace("_", " ").title()
graph += geom_density(alpha=.5)
graph += scale_fill_manual(values=[colors.FIRST_COLOR, colors.SECOND_COLOR],
name=fill_label)
graph += (theme_bw()
+ theme(
axis_line_x=element_line(color='gray'),
axis_line_y=element_line(color='gray'),
line=element_line(color='white')
)
)
# labels
graph += xlab(x_label) + ylab("Density")
graph += scale_y_continuous(labels=percent_format()) # custom_format('{:.2f} USD')
line_args = {
"color": colors.THIRD_COLOR,
"size": .5
}
var_describe = df[variable].describe()
if mean:
graph += geom_vline(xintercept=var_describe.loc["mean"],
linetype="dashed", **line_args
)
if mode:
graph += geom_vline(xintercept=var_describe.loc["50%"],
linetype="solid",
**line_args)
if q1:
graph += geom_vline(xintercept=var_describe.loc["25%"],
linetype="solid",
**line_args)
if q3:
graph += geom_vline(xintercept=var_describe.loc["75%"],
linetype="solid",
**line_args)
# title
if not title is None:
graph += ggtitle(str(title))
# no outliers
if no_outliers:
max_75 = df[variable].describe().loc['75%']
min_25 = df[variable].describe().loc['25%']
graph += xlim(min_25, max_75)
# show
graph.draw()
plt.show()
scale_x_datetime(date_breaks='5 years', date_labels='%Y') +
scale_color_discrete(name='HPI', labels=['CREA', 'Teranet']))
gg_save('gg_tera_crea_lvl.png', dir_figures, gg_tera_crea_lvl, 12, 5)
# (iii) CREA vs Teranet m/m
tmp = df_hpi_both.groupby(
['city',
'hpi']).apply(lambda x: x.value / x.value.shift(1) - 1).reset_index()
df_hpi_w = df_hpi_both.assign(mm=tmp.sort_values('level_2').value.values)
df_hpi_w = df_hpi_w.pivot_table('mm', ['date', 'city'], 'hpi').reset_index()
df_hpi_w = df_hpi_w.sort_values(['city', 'date']).reset_index(None, True)
gg_tera_crea_pct = (ggplot(df_hpi_w, aes(x='crea', y='tera')) +
geom_point(size=0.5) + theme_bw() +
theme(axis_text_x=element_text(angle=90)) +
scale_x_continuous(labels=percent_format()) +
scale_y_continuous(labels=percent_format()) +
labs(x='CREA', y='Teranet', title='month-over-month %') +
facet_wrap('~city', nrow=2))
gg_save('gg_tera_crea_pct.png', dir_figures, gg_tera_crea_pct, 12, 5)
# (iv) Find the optimal correlation
lag_seq = np.arange(13)
alpha = 0.1
n_bs = 1000
holder = []
for lag in lag_seq:
print(lag)
tmp_lag = df_hpi_w.assign(
crea=df_hpi_w.groupby('city').crea.shift(lag)).dropna()
tmp_lag_bs = tmp_lag.groupby('city').sample(frac=n_bs,
