def mon_boxplot(ts,
col='all',
dtype='flow',
fun='mean',
min_yrs=10,
export_path='',
export_name='flow_boxplot.png'):
"""
"""
from os import path
from calendar import month_abbr
import seaborn as sns
import matplotlib.pyplot as plt
from core.ts import w_resample
from core.ts.sw import flow_stats
from pandas import Series
ylabel_dict = {
'flow': 'Mean Flow $(m^{3}/s)$',
'precip': 'Precipitation $(mm/month)$',
'PET': 'Potential Evapotranspiration $(mm/month)$',
'diff': 'Precipitation - PET $(mm/month)$'
}
sns.set_style("whitegrid")
sns.set_context('poster')
col_dict = {
'flow': sns.color_palette("pastel")[0],
'precip': sns.color_palette("pastel")[0],
'PET': sns.color_palette("pastel")[1],
'diff': sns.light_palette("blue")[0]
}
color = col_dict[dtype]
if type(ts) is Series:
ts2 = ts
else:
stats1 = flow_stats(ts)
index1 = stats1.loc['Tot data yrs', :] >= min_yrs
sites = ts.columns[index1]
ts2 = ts[sites]
tsm = w_resample(ts2, period='month', fun=fun)
tsm['month_num'] = tsm.index.month
tsm['month'] = tsm['month_num'].apply(lambda x: month_abbr[x])
if type(ts) is Series:
fig, ax = plt.subplots(figsize=(15, 10))
sns.boxplot(x='month', y=0, data=tsm, color=color, showfliers=False)
plt.ylabel(ylabel_dict[dtype])
plt.xlabel('Month')
plt.tight_layout()
plot2 = ax.get_figure()
plot2.savefig(path.join(export_path, export_name))
elif col is 'all':
for i in sites:
fig, ax = plt.subplots(figsize=(15, 10))
sns.boxplot(x='month',
y=i,
data=tsm,
color=color,
showfliers=False)
plt.ylabel(ylabel_dict[dtype])
plt.xlabel('Month')
plt.tight_layout()
plot2 = ax.get_figure()
plot2.savefig(path.join(export_path, str(i) + '_' + export_name))
elif type(col) is list:
for i in col:
fig, ax = plt.subplots(figsize=(15, 10))
sns.boxplot(x='month', y=i, color=color, data=tsm)
plt.ylabel(ylabel_dict[dtype])
plt.xlabel('Month')
plt.tight_layout()
plot2 = ax.get_figure()
plot2.savefig(path.join(export_path, str(i) + '_' + export_name))
def hydrograph_plot(flow,
precip=None,
x_period='day',
x_n_periods=1,
time_format='%d-%m %H',
x_rot=45,
alpha=0.6,
flow_units='m^{3}/s',
precip_units='mm',
x_grid=False,
export_path=None):
"""
Plotting function for hydrographs with/without a precipitation data set.
Arguments:\n
flow -- Pandas Dataframe time series with sites as columns.\n
precip -- Pandas Dataframe time series with sites as columns.\n
x_period -- The gross period for the axis.\n
x_n_period -- The number of intervals per x_period.\n
time_format -- The Date-time format for the x-axis in ISO format.\n
x_rot -- The rotation of the x-axis labels.\n
alpha -- The transparency of the precip plots.\n
flow_units -- The units that will be shown on the flow y-axis.\n
precip_units -- The units that will be shown on the precip y-axis.\n
x_grid -- Should a vertical grid be drawn?
"""
import matplotlib.pyplot as plt
from numpy import arange, in1d, where
from seaborn import set_style, despine, set_context, color_palette
from core.ts import w_resample
set_style("white")
set_style("ticks")
set_context('poster')
pal1 = color_palette()
#### Set up index and x scales
x_index = flow.index
x_index2 = w_resample(flow, x_period, x_n_periods).index
x_axis = arange(0, len(flow.index))
end = x_axis[-1]
xticks_num = where(in1d(x_index, x_index2))[0]
xticks = x_index[in1d(x_index, x_index2)].strftime(time_format)
#### Make the plots
u1 = flow.reset_index().iloc[:, 1:]
ax1 = u1.plot(x_axis, rot=x_rot)
if precip is not None:
ax2 = ax1.twinx()
u2 = precip.reset_index().iloc[:, 1:]
x_axis2 = x_axis[in1d(flow.index, precip.index)]
precip_names = u2.columns
bar_width = ((x_axis2[1] - x_axis2[0]) * 0.75) / len(precip_names)
for i in arange(0, len(precip_names)):
ax2.bar(x_axis2 + bar_width * i,
u2.iloc[:, i],
bar_width,
color=pal1[i],
label=precip_names[i],
alpha=alpha)
ax2.set_ylabel('Precipitation $(' + precip_units + ')$')
ax2.legend(loc='upper right')
despine(right=False)
else:
despine()
#### Adjust the axes and legends
ax1.xaxis.set_ticks(xticks_num)
plt.xticks(xticks_num, xticks)
plt.xlim(0, end)
ax1.set_ylabel('Flow $(' + flow_units + ')$')
ax1.legend(loc='upper left')
if x_grid:
ax1.xaxis.grid(True, linewidth=0.5, linestyle='--')
despine(right=False, top=False)
plt.tight_layout()
plot = ax1.get_figure()
#### Save plot
if isinstance(export_path, str):
plot.savefig(export_path)
return (plot)
def multi_yr_barplot(ts1,
ts2,
col='all',
names_dict=None,
dtype='flow',
fun='mean',
single=False,
start='1970',
end='2015',
alf=False,
export_path='',
export_name='flow_multi_barplot.png'):
"""
"""
from os import path
from numpy import repeat
from pandas import concat, melt, DatetimeIndex
import seaborn as sns
import matplotlib.pyplot as plt
from core.ts import w_resample, tsreg
from core.ts.sw import flow_stats, malf7d
### Set base parameters
ylabel_dict = {
'flow': 'Mean Flow $(m^{3}/s)$',
'precip': 'Precipitation $(mm/year)$',
'ET': 'Evapotranspiration $(mm/year)$',
'both': '$mm/year$'
}
sns.set_style("whitegrid")
sns.set_context('poster')
cat_dict = {
'flow': ['Recorded', 'Naturalised'],
'both': ['Precipitation', 'Potential Evapotranspiration']
}
cat = cat_dict[dtype]
### assign names if needed
if type(names_dict) is dict:
col_names = [names_dict[x] for x in col]
### Process data
if col is not 'all':
ts1_sel = ts1[col]
ts2_sel = ts2[col]
ts1_sel.columns = ts2_sel.columns = col_names
else:
ts1_sel = ts1
ts2_sel = ts2
ts1_yr = w_resample(ts1_sel, period='water year', fun=fun)
ts2_yr = w_resample(ts2_sel, period='water year', fun=fun)
if alf:
malf1, ts1_yr, alf_mis1 = malf7d(ts1_sel)
malf1, ts2_yr, alf_mis1 = malf7d(ts2_sel)
ts_all = tsreg(
concat([ts1_yr, ts2_yr], axis=1, join='inner',
keys=cat).dropna(how='all'))
ts_all = ts_all[start:end]
if single:
ts_non = ts_all.unstack().reset_index().drop('level_1', axis=1)
time_col = ts_all.index.name
ts_non['year'] = DatetimeIndex(ts_non[time_col]).year
ts_non.columns = ['site', time_col, 'value', 'year']
ts_non.site = ts_non.site.astype('str')
fig, ax = plt.subplots(figsize=(15, 10))
sns.barplot(x='year',
y='value',
hue='site',
data=ts_non,
palette='pastel')
lgd = plt.legend(loc='best')
else:
ts_non = ts_all[cat[0]].unstack().reset_index()
time_col = ts_all.index.name
ts_non['year'] = DatetimeIndex(ts_non[time_col]).year
ts_non.columns = ['site', time_col, 'value', 'year']
ts_non.site = ts_non.site.astype('str')
ts_nat = ts_all[cat[1]].unstack().reset_index()
ts_nat['year'] = DatetimeIndex(ts_nat[time_col]).year
ts_nat.columns = ['site', time_col, 'value', 'year']
ts_nat.site = ts_nat.site.astype('str')
order1 = ts1_sel.columns.astype('str').tolist()
### Make the figures
fig, ax = plt.subplots(figsize=(15, 10))
sns.barplot(x='year',
y='value',
hue='site',
data=ts_nat,
palette='pastel',
hue_order=order1)
sns.barplot(x='year',
y='value',
hue='site',
data=ts_non,
saturation=0.75,
hue_order=order1)
plt.plot([0], marker='None', linestyle='None', label='dummy')
handles, lbs = ax.get_legend_handles_labels()
handles2 = [handles[0]
] + handles[len(col) + 1:] + handles[:len(col) + 1]
labels2 = [cat[0]] + order1 + [cat[1]] + order1
lgd = plt.legend(handles2,
labels2,
loc='center left',
bbox_to_anchor=(1, 0.5))
xticks = ax.get_xticks()
if len(xticks) > 15:
for label in ax.get_xticklabels()[::2]:
label.set_visible(False)
ax.xaxis_date()
fig.autofmt_xdate(ha='center')
# plt.tight_layout()
# ax.xaxis_date()
# fig.autofmt_xdate(ha='center')
if alf:
plt.ylabel('7DALF $(m^{3}/s)$')
else:
plt.ylabel(ylabel_dict[dtype])
plt.xlabel('Water Year')
# lgd = plt.legend([handles[i] for i in order2], [r'$D_{etc}$'] + cat + [r'$A_{etc}$'] + cat, loc='center left', bbox_to_anchor=(1, 0.5), ncol=2)
# plt.tight_layout()
plot2 = ax.get_figure()
plot2.savefig(path.join(export_path, export_name),
bbox_extra_artists=(lgd, ),
bbox_inches='tight')
def reg_plot(x,
y,
freq='day',
n_periods=1,
fun='mean',
min_ratio=0.75,
digits=3,
x_max=None,
y_max=None,
logs=False,
export=False,
export_path='flow_reg.png'):
"""
Function to plot regressions with confidence intervals.
"""
from os import path
from numpy import nan, log
from core.stats.reg import lin_reg
import seaborn as sns
from core.ts.ts import w_resample
from pandas import concat
import matplotlib.pyplot as plt
### Resample and combine
x1 = w_resample(x, freq, n_periods, fun, min_ratio, digits)
y1 = w_resample(y, freq, n_periods, fun, min_ratio, digits)
xy1 = concat([x1, y1], axis=1, join='inner').dropna()
col_names = [str(x.columns[0]), str(y.columns[0])]
xy1.columns = col_names
if logs:
xy1 = xy1.apply(log)
### Make regression
if x_max and y_max is not None:
xy2 = xy1.copy()
xy2.ix[xy2.iloc[:, 0] > x_max, 0] = nan
xy2.ix[xy2.iloc[:, 1] > y_max, 1] = nan
xy2 = xy2.dropna()
reg1 = lin_reg(xy2.iloc[:, 0], xy2.iloc[:, 1])[0]
else:
xy2 = xy1.copy()
reg1 = lin_reg(xy2.iloc[:, 0], xy2.iloc[:, 1])[0]
### plot regression
sns.set_style("whitegrid")
sns.set_context('poster')
fig, ax = plt.subplots(figsize=(15, 10))
# if log_x:
# ax.set(xscale="log")
# if log_y:
# ax.set(yscale="log")
sns.regplot(x=col_names[0],
y=col_names[1],
data=xy2.reset_index(),
truncate=False,
fit_reg=True,
ax=ax)
x_lim = ax.xaxis.get_view_interval()
y_lim = ax.yaxis.get_view_interval()
if x_max is not None:
ax.set_xlim(0, x_max)
# else:
# ax.set_xlim(0, x_lim[1])
if y_max is not None:
ax.set_ylim(0, y_max)
# else:
# ax.set_ylim(0, y_lim[1])
plt.xlabel(col_names[0] + ' $(m^{3}/s)$')
plt.ylabel(col_names[1] + ' $(m^{3}/s)$')
plot2 = ax.get_figure()
if export:
plot2.savefig(path.join(export_path), bbox_inches='tight')
return (plot2, reg1)
def dual_mon_boxplot(ts1,
ts2,
col='all',
dtype='flow',
fun='mean',
min_yrs=10,
export_path='',
export_name='flow_dual_boxplot.png'):
"""
"""
from calendar import month_abbr
from os import path
from pandas import concat, melt, DatetimeIndex
import seaborn as sns
import matplotlib.pyplot as plt
from core.ts import w_resample
from core.ts.sw import flow_stats
### Set base parameters
ylabel_dict = {
'flow': 'Mean Flow $(m^{3}/s)$',
'precip': 'Precipitation $(mm/month)$',
'ET': 'Evapotranspiration $(mm/month)$'
}
sns.set_style("whitegrid")
sns.set_context('poster')
cat = ['Recorded', 'Naturalised']
### Process data
ts1 = ts1.dropna(axis=1, how='all')
stats1 = flow_stats(ts1)
index1 = stats1.loc[:, 'Tot data yrs'] >= min_yrs
sites = ts1.columns[index1.values]
ts1_sel = ts1[sites]
ts2_sel = ts2[sites]
ts1_selm = w_resample(ts1_sel, period='month', fun=fun)
ts2_selm = w_resample(ts2_sel, period='month', fun=fun)
ts_all = concat([ts1_selm, ts2_selm], axis=1, join='inner')
### Make the figures
if col is 'all':
for i in sites:
ts_site = ts_all[i]
ts_site.columns = cat
ts_site2 = melt(ts_site.reset_index(),
id_vars='Time',
value_vars=cat,
var_name='type')
ts_site2['month_num'] = DatetimeIndex(ts_site2.Time).month
ts_site2['month'] = ts_site2['month_num'].apply(
lambda x: month_abbr[x])
ts_site2 = ts_site2.sort_values(['month_num', 'type'],
ascending=[True, False],
axis='index')
fig, ax = plt.subplots(figsize=(15, 10))
sns.boxplot(x='month',
y='value',
hue='type',
data=ts_site2,
showfliers=False)
plt.ylabel(ylabel_dict[dtype])
plt.xlabel('Month')
plt.legend()
plt.tight_layout()
plot2 = ax.get_figure()
plot2.savefig(path.join(export_path, str(i) + '_' + export_name))
elif type(col) is list:
for i in col:
ts_site = ts_all[i]
ts_site.columns = cat
ts_site2 = melt(ts_site.reset_index(),
id_vars='Time',
value_vars=cat,
var_name='type')
ts_site2['month_num'] = DatetimeIndex(ts_site2.Time).month
ts_site2['month'] = ts_site2['month_num'].apply(
lambda x: month_abbr[x])
ts_site2 = ts_site2.sort_values(['month_num', 'type'],
ascending=[True, False],
axis='index')
fig, ax = plt.subplots(figsize=(15, 10))
sns.boxplot(x='month',
y='value',
hue='type',
data=ts_site2,
showfliers=False)
plt.ylabel(ylabel_dict[dtype])
plt.xlabel('Month')
plt.legend()
plt.tight_layout()
plot2 = ax.get_figure()
plot2.savefig(path.join(export_path, str(i) + '_' + export_name))
