def visualize_nodes(labels: list, pca_2d: list, num_of_clusters: int,
features: list, albums: list, task_name: str, method: str):
color_dic = dict()
for i in range(-1, num_of_clusters + 1):
color_dic[i] = random_color()
cls = [color_dic[label] for label in labels]
TOOLTIPS = [('title', '@title'), ('URL', '@url'), ('Cluster', '@cluster')]
source = ColumnDataSource(
data=dict(x=pca_2d[:, 0],
y=pca_2d[:, 1],
title=[album['title'] for album in albums],
url=[album['url'] for album in albums],
color=cls,
cluster=labels))
for feature in features:
TOOLTIPS.append((feature, "@{0}".format(feature)))
source.add([album[feature] for album in albums], feature)
p = figure(sizing_mode='stretch_both',
title=method + " clustering of albums on metrics: " +
", ".join(features),
tooltips=TOOLTIPS,
output_backend='webgl')
p.circle('x', 'y', source=source, color='color', fill_alpha=0.2, size=5)
output_file(get_local_data_path(task_name + ".html"),
title="PSZ | {0} clustering of Albums".format(method))
show(p)
def ridgeplot(courses_obj):
# first format 'courses_obj' into 'probly' DataFrame format
# courses_obj: [{'course_name': 'Calculus...', ...}, ...]
grades = [[100*y[2]/y[3] for y in x['assignments']] for x in courses_obj]
# turn this list of lists into a complete NumPy array
length = len(sorted(grades, key=len, reverse=True)[0])
grades = np.array([xi+[None]*(length-len(xi)) for xi in grades], dtype='float')
columns = [x['course_name'] for x in courses_obj]
grades = grades.transpose()
probly = pd.DataFrame(grades, columns=columns)
cats = list(reversed(probly.keys()))
palette = [cc.rainbow[i*15] for i in range(17)]
x = np.linspace(-20,110, 500)
source = ColumnDataSource(data=dict(x=x))
p = figure(y_range=cats, plot_width=900, plot_height = 300, x_range=(-5, 120))#, toolbar_location=None)
for i, cat in enumerate(reversed(cats)):
adjusted = probly[cat].replace([np.inf, -np.inf], np.nan).dropna(how="all")
if adjusted.size == 1 or pd.unique(adjusted).size == 1: # this means we can't compute
continue
pdf = gaussian_kde(adjusted)
#p = figure(plot_width=400, plot_height=400)
#p.line(x, pdf(x))
y = ridge(cat, pdf(x), scale=2)
#p.line(x, y, color='black')
#show(p)
source.add(y, cat)
p.patch('x', cat, color=palette[i], alpha=0.6, line_color="black", source=source)
p.outline_line_color = None
p.background_fill_color = "#efefef"
p.tools = [PanTool(), CrosshairTool(), HoverTool(), SaveTool(), BoxZoomTool(), WheelZoomTool(), ResetTool()]
ticks = list(np.array([np.array([0,3,7])+i*10 for i in range(10)]).flatten()) + [100]
p.xaxis.ticker = FixedTicker(ticks=ticks)
p.xaxis.formatter = PrintfTickFormatter(format="%d")
p.xaxis.axis_label = "Your Grade Distribution"
p.yaxis.axis_label = "Your Courses"
p.ygrid.grid_line_color = None
p.xgrid.grid_line_color = "Grey"
p.xgrid.ticker = p.xaxis[0].ticker
p.axis.minor_tick_line_color = None
p.axis.major_tick_line_color = None
p.axis.axis_line_color = None
p.y_range.range_padding = 0.12
return p
def pnl_chart(df):
"""
Creates a chart to display P&L
Parameters
----------
df : DataFrame, with the following columns: Date, Realized, Unrealized, Total
Returns
-------
Figure
"""
source = ColumnDataSource(df)
source.add(df.index.map(lambda d: d.strftime('%d-%b-%y')), 'DateStr')
title = Title(text='Historical Daily P&L',
text_font_size='14pt',
align='center')
p = figure(x_axis_type='datetime', sizing_mode='scale_height', title=title)
p.xaxis.formatter = DatetimeTickFormatter(days=['%d-%b-%y'],
months=['%d-%b-%y'],
years=['%d-%b-%y'])
p.xaxis.major_label_text_font_size = '10pt'
p.yaxis.formatter = NumeralTickFormatter(format=',')
p.yaxis.major_label_text_font_size = '10pt'
for col, color in zip(['Realized', 'Unrealized', 'Total'], Spectral4):
p.line(x='Date',
y=col,
source=source,
line_width=2,
color=color,
alpha=0.8,
legend_label=col)
hover = HoverTool(tooltips=[
('Date', '@DateStr'),
('Realized', '@Realized{,}'),
('Unrealized', '@Unrealized{,}'),
('Total', '@Total{,}'),
],
formatters={
'Realized': 'printf',
'Unrealized': 'printf',
'Total': 'printf'
},
mode='vline',
renderers=[p.renderers[-1]])
p.add_tools(hover)
p.legend.location = "top_left"
p.legend.click_policy = "hide"
return p
def _to_cds(data, unstack=None):
""""""
source_dict = {}
if hasattr(data, "data_vars"):
for v in data.data_vars:
source_dict.update(_dict_from_da(data[v], unstack))
else:
source_dict.update(_dict_from_da(data, unstack))
source = ColumnDataSource(source_dict)
for c in data.coords:
if c not in data.dims and data[c].ndim == 1:
source.add(data[c].values, c)
return source
class PlotterBokeh(Plotter):
def __init__(self, ds, x, y, z=None, **kwargs):
"""
"""
# bokeh custom options / defaults
kwargs['return_fig'] = kwargs.pop('return_fig', False)
self._interactive = kwargs.pop('interactive', False)
super().__init__(ds, x, y, z, **kwargs, backend='BOKEH')
def prepare_axes(self):
"""Make the bokeh plot figure and set options.
"""
from bokeh.plotting import figure
if self.add_to_axes is not None:
self._plot = self.add_to_axes
else:
# Currently axes scale type must be set at figure creation?
self._plot = figure(
# convert figsize to roughly matplotlib dimensions
width=int(self.figsize[0] * 80 +
(100 if self._use_legend else 0) +
(20 if self._ytitle else 0) +
(20 if not self.yticklabels_hide else 0)),
height=int(self.figsize[1] * 80 + (20 if self.title else 0) +
(20 if self._xtitle else 0) +
(20 if not self.xticklabels_hide else 0)),
x_axis_type=('log' if self.xlog else 'linear'),
y_axis_type=('log' if self.ylog else 'linear'),
y_axis_location=('right' if self.ytitle_right else 'left'),
title=self.title,
toolbar_location="above",
toolbar_sticky=False,
active_scroll="wheel_zoom",
)
def set_axes_labels(self):
"""Set the labels on the axes.
"""
if self._xtitle:
self._plot.xaxis.axis_label = self._xtitle
if self._ytitle:
self._plot.yaxis.axis_label = self._ytitle
def set_axes_range(self):
"""Set the plot ranges of the axes, and the panning limits.
"""
from bokeh.models import DataRange1d
self.calc_data_range()
# plt_x_centre = (self._data_xmax + self._data_xmin) / 2
# plt_x_range = self._data_xmax - self._data_xmin
# xbounds = (plt_x_centre - plt_x_range, plt_x_centre + plt_x_range)
xbounds = None
self._plot.x_range = (DataRange1d(
start=self._xlims[0], end=self._xlims[1], bounds=xbounds)
if self._xlims else DataRange1d(bounds=xbounds))
# plt_y_centre = (self._data_ymax + self._data_ymin) / 2
# plt_y_range = abs(self._data_ymax - self._data_ymin)
# ybounds = (plt_y_centre - plt_y_range, plt_y_centre + plt_y_range)
ybounds = None
self._plot.y_range = (DataRange1d(
start=self._ylims[0], end=self._ylims[1], bounds=ybounds)
if self._ylims else DataRange1d(bounds=ybounds))
def set_spans(self):
"""Set custom horizontal and verical line spans.
"""
from bokeh.models import Span
span_opts = {
'level': 'glyph',
'line_dash': 'dashed',
'line_color': (127, 127, 127),
'line_width': self.span_width,
}
if self.hlines:
for hl in self.hlines:
self._plot.add_layout(
Span(location=hl, dimension='width', **span_opts))
if self.vlines:
for vl in self.vlines:
self._plot.add_layout(
Span(location=vl, dimension='height', **span_opts))
def set_gridlines(self):
"""Set whether to use gridlines or not.
"""
if not self.gridlines:
self._plot.xgrid.visible = False
self._plot.ygrid.visible = False
else:
self._plot.xgrid.grid_line_dash = self.gridline_style
self._plot.ygrid.grid_line_dash = self.gridline_style
def set_tick_marks(self):
"""Set custom locations for the tick marks.
"""
from bokeh.models import FixedTicker
if self.xticks:
self._plot.xaxis[0].ticker = FixedTicker(ticks=self.xticks)
if self.yticks:
self._plot.yaxis[0].ticker = FixedTicker(ticks=self.yticks)
if self.xticklabels_hide:
self._plot.xaxis.major_label_text_font_size = '0pt'
if self.yticklabels_hide:
self._plot.yaxis.major_label_text_font_size = '0pt'
def set_sources_heatmap(self):
from bokeh.plotting import ColumnDataSource
# initialize empty source
if not hasattr(self, '_source'):
self._source = ColumnDataSource(data=dict())
# remove mask from data -> not necessary soon? / convert to nan?
var = np.ma.getdata(self._heatmap_var)
self._source.add([var], 'image')
self._source.add([self._data_xmin], 'x')
self._source.add([self._data_ymin], 'y')
self._source.add([self._data_xmax - self._data_xmin], 'dw')
self._source.add([self._data_ymax - self._data_ymin], 'dh')
def set_sources(self):
"""Set the source dictionaries to be used by the plotter functions.
This is seperate to allow interactive updates of the data only.
"""
from bokeh.plotting import ColumnDataSource
# check if heatmap
if hasattr(self, '_heatmap_var'):
return self.set_sources_heatmap()
# 'copy' the zlabels iterator into src_zlbs
self._zlbls, src_zlbs = itertools.tee(self._zlbls)
# Initialise with empty dicts
if not hasattr(self, "_sources"):
self._sources = [
ColumnDataSource(dict()) for _ in range(len(self._z_vals))
]
# range through all data and update the sources
for i, (zlabel, data) in enumerate(zip(src_zlbs, self._gen_xy())):
self._sources[i].add(data['x'], 'x')
self._sources[i].add(data['y'], 'y')
self._sources[i].add([zlabel] * len(data['x']), 'z_coo')
# check for color for scatter plot
if 'c' in data:
self._sources[i].add(data['c'], 'c')
# check if should set y_err as well
if 'ye' in data:
y_err_p = data['y'] + data['ye']
y_err_m = data['y'] - data['ye']
self._sources[i].add(list(zip(data['x'], data['x'])),
'y_err_xs')
self._sources[i].add(list(zip(y_err_p, y_err_m)), 'y_err_ys')
# check if should set x_err as well
if 'xe' in data:
x_err_p = data['x'] + data['xe']
x_err_m = data['x'] - data['xe']
self._sources[i].add(list(zip(data['y'], data['y'])),
'x_err_ys')
self._sources[i].add(list(zip(x_err_p, x_err_m)), 'x_err_xs')
def plot_legend(self, legend_items=None):
"""Add a legend to the plot.
"""
if self._use_legend:
from bokeh.models import Legend
loc = {'best': 'top_left'}.get(self.legend_loc, self.legend_loc)
where = {None: 'right'}.get(self.legend_where, self.legend_where)
# might be manually specified, e.g. from multiplot
if legend_items is None:
legend_items = self._lgnd_items
lg = Legend(items=legend_items)
lg.location = loc
lg.click_policy = 'hide'
self._plot.add_layout(lg, where)
# Don't repeatedly redraw legend
self._use_legend = False
def set_mappable(self):
from bokeh.models import LogColorMapper, LinearColorMapper
import matplotlib as plt
mappr_fn = (LogColorMapper if self.colormap_log else LinearColorMapper)
bokehpalette = [plt.colors.rgb2hex(m) for m in self.cmap(range(256))]
self.mappable = mappr_fn(palette=bokehpalette,
low=self._zmin,
high=self._zmax)
def plot_colorbar(self):
if self._use_colorbar:
where = {None: 'right'}.get(self.legend_where, self.legend_where)
from bokeh.models import ColorBar, LogTicker, BasicTicker
ticker = LogTicker if self.colormap_log else BasicTicker
color_bar = ColorBar(color_mapper=self.mappable,
location=(0, 0),
ticker=ticker(desired_num_ticks=6),
title=self._ctitle)
self._plot.add_layout(color_bar, where)
def set_tools(self):
"""Set which tools appear for the plot.
"""
from bokeh.models import HoverTool
self._plot.add_tools(
HoverTool(tooltips=[("({}, {})".format(
self.x_coo,
self.y_coo if isinstance(self.y_coo, str) else None),
"(@x, @y)"), (self.z_coo, "@z_coo")]))
def update(self):
from bokeh.io import push_notebook
self.set_sources()
push_notebook()
def show(self, **kwargs):
"""Show the produced figure.
"""
if self.return_fig:
return self._plot
bshow(self._plot, **kwargs)
return self
def prepare_plot(self):
self.prepare_axes()
self.set_axes_labels()
self.set_axes_range()
self.set_spans()
self.set_gridlines()
self.set_tick_marks()
self.set_sources()
from bokeh.plotting import ColumnDataSource, figure, gridplot, output_file, show
from bokeh.sampledata.autompg import autompg
output_file("panning.html")
# Load some Automobile data into a data source. Interesting columns are:
# "yr" - Year manufactured
# "mpg" - miles per gallon
# "displ" - engine displacement
# "hp" - engine horsepower
# "cyl" - number of cylinders
source = ColumnDataSource(autompg.to_dict("list"))
source.add(autompg["yr"], name="yr")
# Let's set up some plot options in a dict that we can re-use on multiple plots
plot_config = dict(plot_width=300,
plot_height=300,
tools="pan,wheel_zoom,box_zoom,select")
# First let's plot the "yr" vs "mpg" using the plot config above
# Note that we are supplying our our data source to the renderer explicitly
p1 = figure(title="MPG by Year", **plot_config)
p1.circle("yr", "mpg", color="blue", source=source)
# EXERCISE: make another figure p2 with circle renderer, for "hp" vs "displ" with
# color "green". This renderer should use the same data source as the renderer
# above, that is what will cause the plots selections to be linked
p2 = figure(title="HP vs. Displacement", **plot_config)
p2.circle("hp", "displ", color="green", source=source)
# EXERCISE: and another figure p3 with circle renderer for "mpg" vs "displ",
from bokeh.plotting import ColumnDataSource, figure, gridplot, output_file, show
from bokeh.sampledata.autompg import autompg
output_file("brushing.html")
# Load some Automobile data into a data source. Interesting columns are:
# "yr" - Year manufactured
# "mpg" - miles per gallon
# "displ" - engine displacement
# "hp" - engine horsepower
# "cyl" - number of cylinders
source = ColumnDataSource(autompg.to_dict("list"))
source.add(autompg["yr"], name="yr")
# define some tools to add
TOOLS = "pan,wheel_zoom,box_zoom,box_select,lasso_select"# Let's set up some plot options in a dict that we can re-use on multiple plots
# Let's set up some plot options in a dict that we can re-use on multiple plots
plot_config = dict(plot_width=300, plot_height=300, tools=TOOLS)
# First let's plot the "yr" vs "mpg" using the plot config above
# Note that we are supplying our our data source to the renderer explicitly
p1 = figure(title="MPG by Year", **plot_config)
p1.circle("yr", "mpg", color="blue", source=source)
# EXERCISE: make another figure p2 with circle renderer, for "hp" vs "displ" with
# color "green". This renderer should use the same data source as the renderer
# above, that is what will cause the plots selections to be linked
p2 = figure(title="HP vs. Displacement", **plot_config)
p2.circle("hp", "displ", color="green", source=source)
def _plot_get_source(self, conf_list, runs, X, inc_list, hp_names):
"""
Create ColumnDataSource with all the necessary data
Contains for each configuration evaluated on any run:
- all parameters and values
- origin (if conflicting, origin from best run counts)
- type (default, incumbent or candidate)
- # of runs
- size
- color
Parameters
----------
conf_list: list[Configuration]
configurations
runs: list[int]
runs per configuration (same order as conf_list)
X: np.array
configuration-parameters as 2-dimensional array
inc_list: list[Configuration]
incumbents for this conf-run
hp_names: list[str]
names of hyperparameters
Returns
-------
source: ColumnDataSource
source with attributes as requested
"""
# Remove all configurations without any runs
keep = [i for i in range(len(runs)) if runs[i] > 0]
runs = np.array(runs)[keep]
conf_list = np.array(conf_list)[keep]
X = X[keep]
source = ColumnDataSource(data=dict(x=X[:, 0], y=X[:, 1]))
for k in hp_names: # Add parameters for each config
source.add([c[k] if c[k] else "None" for c in conf_list],
escape_parameter_name(k))
default = conf_list[0].configuration_space.get_default_configuration()
conf_types = [
"Default" if c == default else "Final Incumbent" if c
== inc_list[-1] else "Incumbent" if c in inc_list else "Candidate"
for c in conf_list
]
# We group "Local Search" and "Random Search (sorted)" both into local
origins = [self._get_config_origin(c) for c in conf_list]
source.add(conf_types, 'type')
source.add(origins, 'origin')
sizes = self._get_size(runs)
sizes = [
s * 3 if conf_types[idx] == "Default" else s
for idx, s in enumerate(sizes)
]
source.add(sizes, 'size')
source.add(self._get_color(source.data['type']), 'color')
source.add(runs, 'runs')
# To enforce zorder, we categorize all entries according to their size
# Since we plot all different zorder-levels sequentially, we use a
# manually defined level of influence
num_bins = 20 # How fine-grained the size-ordering should be
min_size, max_size = min(source.data['size']), max(source.data['size'])
step_size = (max_size - min_size) / num_bins
if step_size == 0:
step_size = 1
zorder = [
str(int((s - min_size) / step_size)) for s in source.data['size']
]
source.add(zorder, 'zorder') # string, so we can apply group filter
return source
def plot_interactive_footprint(self):
"""Use bokeh to create an interactive algorithm footprint with zoom and
hover tooltips. Should avoid problems with overplotting (since we can
zoom) and provide better information about instances."""
features = np.array(self.features_2d)
instances = self.insts
runhistory = self.rh
algo = {v: k for k, v in self.algo_name.items()}
incumbent = algo['incumbent']
default = algo['default']
source = ColumnDataSource(data=dict(x=features[:, 0], y=features[:,
1]))
# Add all necessary information for incumbent and default
source.add(instances, 'instance_name')
instance_set = [
'train' if i in self.train_feats.keys() else 'test'
for i in instances
]
source.add(instance_set, 'instance_set') # train or test
for config, name in [(incumbent, 'incumbent'), (default, 'default')]:
cost = get_cost_dict_for_config(runhistory, config)
source.add([cost[i] for i in instances], '{}_cost'.format(name))
# TODO should be in function
good, bad = self._get_good_bad(config)
color = [
1 if idx in good else 0 for idx, i in enumerate(instances)
]
# TODO end
color = ['blue' if c else 'red' for c in color]
self.logger.debug("%s colors: %s", name, str(color))
source.add(color, '{}_color'.format(name))
source.add(source.data['default_color'], 'color')
# Define what appears in tooltips
hover = HoverTool(tooltips=[
('instance name', '@instance_name'),
('def cost', '@default_cost'),
('inc_cost', '@incumbent_cost'),
('set', '@instance_set'),
])
# Add radio-button
def_inc_callback = CustomJS(args=dict(source=source),
code="""
var data = source.data;
if (cb_obj.active == 0) {
data['color'] = data['default_color'];
} else {
data['color'] = data['incumbent_color'];
}
source.change.emit();
""")
def_inc_radio_button = RadioButtonGroup(
labels=["default", "incumbent"],
active=0,
callback=def_inc_callback)
# Plot
x_range = DataRange1d(bounds='auto',
start=min(features[:, 0]) - 1,
end=max(features[:, 0]) + 1)
y_range = DataRange1d(bounds='auto',
start=min(features[:, 1]) - 1,
end=max(features[:, 1]) + 1)
p = figure(
plot_height=500,
plot_width=600,
tools=[hover, 'save', 'wheel_zoom', 'box_zoom', 'pan', 'reset'],
active_drag='box_zoom',
x_range=x_range,
y_range=y_range)
# Scatter train and test individually to toggle them
train_view = CDSView(
source=source,
filters=[GroupFilter(column_name='instance_set', group='train')])
test_view = CDSView(
source=source,
filters=[GroupFilter(column_name='instance_set', group='test')])
train = p.scatter(x='x',
y='y',
source=source,
view=train_view,
color='color')
test = p.scatter(x='x',
y='y',
source=source,
view=test_view,
color='color')
p.xaxis.axis_label, p.yaxis.axis_label = 'principal component 1', 'principal component 2'
p.xaxis.axis_label_text_font_size = p.yaxis.axis_label_text_font_size = "15pt"
train_test_callback = CustomJS(args=dict(source=source,
train_view=train,
test_view=test),
code="""
var data = source.data;
if (cb_obj.active == 0) {
train_view.visible = true;
test_view.visible = true;
} else if (cb_obj.active == 1) {
train_view.visible = true;
test_view.visible = false;
} else {
train_view.visible = false;
test_view.visible = true;
}
""")
train_test_radio_button = RadioButtonGroup(
labels=["all", "train", "test"],
active=0,
callback=train_test_callback)
# Export and return
if self.output_dir:
path = os.path.join(self.output_dir,
"content/images/algorithm_footprint.png")
export_bokeh(p, path, self.logger)
layout = column(
p,
row(widgetbox(def_inc_radio_button),
widgetbox(train_test_radio_button)))
return layout
from bokeh.models import HoverTool
from bokeh.models.widgets import DateRangeSlider
data = pd.read_csv('Lekagul Sensor Data.csv')
data_car_id = data['car-id'].as_matrix()
data.columns
plot_width = 1100
plot_height = 500
width = 1
hover = HoverTool()
hover.tooltips = [('index', '$index')]
all_gate_names = np.unique(data['gate-name'].as_matrix())
all_car_types = np.unique(data['car-type'].as_matrix())
source = ColumnDataSource()
source.add(all_gate_names, name='Gate Names')
legend_var = [
'2 axle car (or motorcycle)', '2 axle Truck', 'Ranger', '3 axle Truck',
'4 axle (and above) Truck', '2 axle Bus', '3 axle Bus'
]
start_y = '2016'
start_m = '02'
start_d = '10'
end_y = '2016'
end_m = '04'
end_d = '20'
start_mask = '2016-02-10'
end_mask = '2016-04-20'
mask = (data['Timestamp'] >= start_mask) & (data['Timestamp'] <= end_mask)
class PlotterBokeh(Plotter):
def __init__(self, ds, x, y, z=None, **kwargs):
"""
"""
# bokeh custom options / defaults
kwargs['return_fig'] = kwargs.pop('return_fig', False)
self._interactive = kwargs.pop('interactive', False)
super().__init__(ds, x, y, z, **kwargs, backend='BOKEH')
def prepare_axes(self):
"""Make the bokeh plot figure and set options.
"""
from bokeh.plotting import figure
if self.add_to_axes is not None:
self._plot = self.add_to_axes
else:
# Currently axes scale type must be set at figure creation?
self._plot = figure(
# convert figsize to roughly matplotlib dimensions
width=int(self.figsize[0] * 80 +
(100 if self._use_legend else 0) +
(20 if self._ytitle else 0) +
(20 if not self.yticklabels_hide else 0)),
height=int(self.figsize[1] * 80 +
(20 if self.title else 0) +
(20 if self._xtitle else 0) +
(20 if not self.xticklabels_hide else 0)),
x_axis_type=('log' if self.xlog else 'linear'),
y_axis_type=('log' if self.ylog else 'linear'),
y_axis_location=('right' if self.ytitle_right else 'left'),
title=self.title,
toolbar_location="above",
toolbar_sticky=False,
active_scroll="wheel_zoom",
)
def set_axes_labels(self):
"""Set the labels on the axes.
"""
if self._xtitle:
self._plot.xaxis.axis_label = self._xtitle
if self._ytitle:
self._plot.yaxis.axis_label = self._ytitle
def set_axes_range(self):
"""Set the plot ranges of the axes, and the panning limits.
"""
from bokeh.models import DataRange1d
self.calc_data_range()
# plt_x_centre = (self._data_xmax + self._data_xmin) / 2
# plt_x_range = self._data_xmax - self._data_xmin
# xbounds = (plt_x_centre - plt_x_range, plt_x_centre + plt_x_range)
xbounds = None
self._plot.x_range = (DataRange1d(start=self._xlims[0],
end=self._xlims[1],
bounds=xbounds) if self._xlims else
DataRange1d(bounds=xbounds))
# plt_y_centre = (self._data_ymax + self._data_ymin) / 2
# plt_y_range = abs(self._data_ymax - self._data_ymin)
# ybounds = (plt_y_centre - plt_y_range, plt_y_centre + plt_y_range)
ybounds = None
self._plot.y_range = (DataRange1d(start=self._ylims[0],
end=self._ylims[1],
bounds=ybounds) if self._ylims else
DataRange1d(bounds=ybounds))
def set_spans(self):
"""Set custom horizontal and verical line spans.
"""
from bokeh.models import Span
span_opts = {
'level': 'glyph',
'line_dash': 'dashed',
'line_color': (127, 127, 127),
'line_width': self.span_width,
}
if self.hlines:
for hl in self.hlines:
self._plot.add_layout(Span(
location=hl, dimension='width', **span_opts))
if self.vlines:
for vl in self.vlines:
self._plot.add_layout(Span(
location=vl, dimension='height', **span_opts))
def set_gridlines(self):
"""Set whether to use gridlines or not.
"""
if not self.gridlines:
self._plot.xgrid.visible = False
self._plot.ygrid.visible = False
else:
self._plot.xgrid.grid_line_dash = self.gridline_style
self._plot.ygrid.grid_line_dash = self.gridline_style
def set_tick_marks(self):
"""Set custom locations for the tick marks.
"""
from bokeh.models import FixedTicker
if self.xticks:
self._plot.xaxis[0].ticker = FixedTicker(ticks=self.xticks)
if self.yticks:
self._plot.yaxis[0].ticker = FixedTicker(ticks=self.yticks)
if self.xticklabels_hide:
self._plot.xaxis.major_label_text_font_size = '0pt'
if self.yticklabels_hide:
self._plot.yaxis.major_label_text_font_size = '0pt'
def set_sources_heatmap(self):
from bokeh.plotting import ColumnDataSource
# initialize empty source
if not hasattr(self, '_source'):
self._source = ColumnDataSource(data=dict())
# remove mask from data -> not necessary soon? / convert to nan?
var = np.ma.getdata(self._heatmap_var)
self._source.add([var], 'image')
self._source.add([self._data_xmin], 'x')
self._source.add([self._data_ymin], 'y')
self._source.add([self._data_xmax - self._data_xmin], 'dw')
self._source.add([self._data_ymax - self._data_ymin], 'dh')
def set_sources(self):
"""Set the source dictionaries to be used by the plotter functions.
This is seperate to allow interactive updates of the data only.
"""
from bokeh.plotting import ColumnDataSource
# check if heatmap
if hasattr(self, '_heatmap_var'):
return self.set_sources_heatmap()
# 'copy' the zlabels iterator into src_zlbs
self._zlbls, src_zlbs = itertools.tee(self._zlbls)
# Initialise with empty dicts
if not hasattr(self, "_sources"):
self._sources = [ColumnDataSource(dict())
for _ in range(len(self._z_vals))]
# range through all data and update the sources
for i, (zlabel, data) in enumerate(zip(src_zlbs, self._gen_xy())):
self._sources[i].add(data['x'], 'x')
self._sources[i].add(data['y'], 'y')
self._sources[i].add([zlabel] * len(data['x']), 'z_coo')
# check for color for scatter plot
if 'c' in data:
self._sources[i].add(data['c'], 'c')
# check if should set y_err as well
if 'ye' in data:
y_err_p = data['y'] + data['ye']
y_err_m = data['y'] - data['ye']
self._sources[i].add(
list(zip(data['x'], data['x'])), 'y_err_xs')
self._sources[i].add(list(zip(y_err_p, y_err_m)), 'y_err_ys')
# check if should set x_err as well
if 'xe' in data:
x_err_p = data['x'] + data['xe']
x_err_m = data['x'] - data['xe']
self._sources[i].add(
list(zip(data['y'], data['y'])), 'x_err_ys')
self._sources[i].add(list(zip(x_err_p, x_err_m)), 'x_err_xs')
def plot_legend(self, legend_items=None):
"""Add a legend to the plot.
"""
if self._use_legend:
from bokeh.models import Legend
loc = {'best': 'top_left'}.get(self.legend_loc, self.legend_loc)
where = {None: 'right'}.get(self.legend_where, self.legend_where)
# might be manually specified, e.g. from multiplot
if legend_items is None:
legend_items = self._lgnd_items
lg = Legend(items=legend_items)
lg.location = loc
lg.click_policy = 'hide'
self._plot.add_layout(lg, where)
# Don't repeatedly redraw legend
self._use_legend = False
def set_mappable(self):
from bokeh.models import LogColorMapper, LinearColorMapper
import matplotlib as plt
mappr_fn = (LogColorMapper if self.colormap_log else LinearColorMapper)
bokehpalette = [plt.colors.rgb2hex(m) for m in self.cmap(range(256))]
self.mappable = mappr_fn(palette=bokehpalette,
low=self._zmin, high=self._zmax)
def plot_colorbar(self):
if self._use_colorbar:
where = {None: 'right'}.get(self.legend_where, self.legend_where)
from bokeh.models import ColorBar, LogTicker, BasicTicker
ticker = LogTicker if self.colormap_log else BasicTicker
color_bar = ColorBar(color_mapper=self.mappable, location=(0, 0),
ticker=ticker(desired_num_ticks=6),
title=self._ctitle)
self._plot.add_layout(color_bar, where)
def set_tools(self):
"""Set which tools appear for the plot.
"""
from bokeh.models import HoverTool
self._plot.add_tools(HoverTool(tooltips=[
("({}, {})".format(self.x_coo, self.y_coo
if isinstance(self.y_coo, str) else None),
"(@x, @y)"), (self.z_coo, "@z_coo")]))
def update(self):
from bokeh.io import push_notebook
self.set_sources()
push_notebook()
def show(self, **kwargs):
"""Show the produced figure.
"""
if self.return_fig:
return self._plot
bshow(self._plot, **kwargs)
return self
def prepare_plot(self):
self.prepare_axes()
self.set_axes_labels()
self.set_axes_range()
self.set_spans()
self.set_gridlines()
self.set_tick_marks()
self.set_sources()
def _prepare_data(connections, stations, lines, data, radius_feature,
color_feature, features):
"""
Prepares data for plotting, creates objects necessary for bokeh.
Parameters:
-----------
connections, stations, lines - data from metro class
data : pd.DataFrame()
data prepared by metro class
radius_feature : str
feature name to use as radius scaling
color_feature : str
feature name to use as color for nodes
features : list of strings
list of column names to show on graph
Returns:
-------
source : ColumnDataSource with data
TOOLTIPS : Tooltips for plot
graph : networkx graph
locations : dictionary of stations and normalized coordinates
"""
# create graph
stations['node_name'] = stations['name'] + '_' + stations['line'].astype(
str)
connections['time'] = 1
graph = nx.Graph()
for connection_id, connection in connections.iterrows():
ind1 = connection['station1'] - 1
ind2 = connection['station2'] - 1
station1_name = stations.iloc[ind1]['node_name']
station2_name = stations.iloc[ind2]['node_name']
graph.add_edge(station1_name, station2_name, time=connection['time'])
normed = stations[['longitude', 'latitude']]
normed = normed - normed.min()
normed = normed / normed.max()
locations = dict(
zip(stations['node_name'], normed[['longitude', 'latitude']].values))
x = []
y = []
name = []
radius = []
fill_color = []
line_l = []
d = {i: [] for i in features}
for node in graph.nodes():
# main values
x.append(locations[node][0])
y.append(locations[node][1])
name.append(node.split('_')[0])
radius.append(
np.clip(
.01 * data.loc[(data['display_name'] == node.split('_')[0]) &
(data['line'] == int(node.split('_')[1])),
radius_feature].values[0], 0.003, 1))
colour = _pseudocolor(
data.loc[(data['display_name'] == node.split('_')[0]) &
(data['line'] == int(node.split('_')[1])),
color_feature].values[0])
fill_color.append(_rgb2hex(tuple([int(np.round(i)) for i in colour])))
line_l.append(lines.loc[lines.line == int(node.split('_')[1]),
'name'].values[0])
# user defined values
for k in d.keys():
d[k].append(
str(
np.round(
data.loc[(data['display_name'] == node.split('_')[0]) &
(data['line'] == int(node.split('_')[1])),
'total_traffic'].values[0], 2)))
source = ColumnDataSource(data=dict(x=x,
y=y,
radius=radius,
fill_color=fill_color,
name=name,
line_l=line_l))
for k, v in d.items():
source.add(v, k)
TOOLTIPS = []
for i in features:
TOOLTIPS.append((i, f'@{i}'))
return source, TOOLTIPS, graph, locations
def stack_lines(datas, index_to_string, ylabel = "MW"):
keys = sorted(datas.keys())
nb_max_values = max([ len(datas[key]) for key in keys])
indices = range(nb_max_values)
#palette= brewer["Spectral"][nb_max_values]
palette = bokeh.palettes.viridis(len(keys))
colors = {key : palette[i] for i,key in enumerate(keys)}
xrange = [index_to_string(i) for i in range(nb_max_values+1)]
lines = {}
columns = {}
for key in keys :
columns[key] = [ datas[key][i] for i in indices ]
ds = ColumnDataSource(columns)
ds.add([index_to_string(i) for i in range(nb_max_values)], "left")
ds.add([index_to_string(i+1) for i in range(nb_max_values)], "right")
p=figure(x_range = xrange,
height = defaults.height,
width = defaults.width
)
#print(ds.data)
partial_sum = [ 0. for i in range(nb_max_values)]
legend_items={}
for key in keys:
ds.add( partial_sum , "bottom"+str(key))
ds.add([ partial_sum[i] + datas[key][i] for i in range(nb_max_values)], "top"+str(key))
quad = p.quad(
top = "top"+str(key) ,
bottom = "bottom"+str(key),
left = "left",
right = "right",
color = colors[key],
source = ds,
fill_alpha = 0.7,
hover_fill_alpha = 0.99,
hover_fill_color=colors[key],
hover_line_color=colors[key],
)
legend_items[key] = [quad]
partial_sum = ds.data["top"+str(key)]
legend = Legend(items=[
(key, legend_items[key])
for key in legend_items.keys()
], location=(0, 0), orientation="horizontal")
p.add_layout(legend, "below")
p.yaxis.axis_label = ylabel
#if there is two many lines, we do not put the hover tool
if len(keys) > 10:
return p
hover = HoverTool(
tooltips=
[("time", "@left")]
+ [ (key, "@"+key+ " "+ylabel) for key in keys]
, attachment="horizontal"
, show_arrow=True
)
p.add_tools(hover)
return p
def plot_lines(datas_, index_to_string, ylabel = "MW"):
datas = { str(key).replace(" ", "_") : value for key, value in datas_.items()}
keys = sorted(datas.keys())
nb_max_values = max([ len(datas[key]) for key in keys])
indices = range(nb_max_values)
palette = bokeh.palettes.viridis(len(keys))
colors = {key : palette[i] for i,key in enumerate(keys)}
xrange = [index_to_string(i) for i in range(nb_max_values+1)]
columns = {}
for key in keys :
columns[key] = [ datas[key][i] for i in indices ]
lines = {}
lines["time"] = []
for i in indices:
lines["time"]+= [index_to_string(i), index_to_string(i+1)]
for key in keys:
lines[key] = []
for i in indices :
v = datas[key][i]
lines[key] += [v, v]
ds = ColumnDataSource(lines)
#print(xrange)
p=figure(x_range = xrange
, height = defaults.height,
width = defaults.width
)
legend_items={}
lines = []
for key in keys:
line = p.line( "time", key,
color = colors[key],
source = ds,
line_width = 4,
alpha = 0.7
)
legend_items[key] = [line]
lines += [line]
legend = Legend(
items=[ (key, legend_items[key])
for key in sorted(legend_items.keys()) ]
, location=(0, 0)
, orientation="horizontal"
)
p.add_layout(legend, "below")
p.yaxis.axis_label = ylabel
#if there is two many lines, we do not put the hover tool
if len(keys) > 20:
return p
columns = {}
for key in keys :
columns[key] = [ datas[key][i] for i in indices ]
hover_ds = ColumnDataSource(columns)
hover_ds.add([index_to_string(i) for i in range(nb_max_values)], "time")
hover_ds.add([index_to_string(i) for i in range(nb_max_values)], "left")
hover_ds.add([index_to_string(i+1) for i in range(nb_max_values)], "right")
max_val = -math.inf
min_val = math.inf
for key in keys :
for i in indices :
v = datas[key][i]
if math.isfinite(v):
max_val = max(max_val, v)
min_val = min(min_val, v)
hover_ds.add([max_val for i in range(nb_max_values)], "top")
hover_ds.add([min_val for i in range(nb_max_values)], "bottom")
hover_glyphs = []
#for key in keys:
glyph = p.quad(top="top", bottom="bottom",
left="left", right="right",
alpha = 0,
source = hover_ds,
color = "white",
hover_alpha=0.1,
hover_color = "black")
hover_glyphs += [glyph]
hover = HoverTool(
tooltips = [("time", "@time")]
+ [ (key, "@"+key + " " + ylabel ) for key in keys ]
, attachment="horizontal"
#, mode = "vline"
, renderers = hover_glyphs
)
p.add_tools(hover)
return p
def commonPlot(r, ldict, y1lim, dset, height=None, width=None, y2=None):
"""
"""
tools = "pan, wheel_zoom, box_zoom, crosshair, reset, save"
title = ldict['title']
xlabel = ldict['xlabel']
y1label = ldict['y1label']
p = figure(title=title,
x_axis_type='datetime',
x_axis_label=xlabel,
y_axis_label=y1label,
tools=tools,
output_backend="webgl")
if height is not None:
p.plot_height = height
if width is not None:
p.plot_width = width
if y2 is not None:
p.extra_y_ranges = y2
p.add_layout(
LinearAxis(y_range_name="y2", axis_label=ldict['y2label']),
'right')
# Annoyingly, the main y-axis still autoscales if there is a
# second y-axis. Setting these means that the axis WON'T
# autoscale until they're set back to None
p.y_range = DataRange1d(start=y1lim[0], end=y1lim[1])
p.x_range.follow = "end"
p.x_range.range_padding = 0.1
p.x_range.range_padding_units = 'percent'
# Hack! But it works. Need to do this *before* you create cds below!
# Includes a special flag (first=True) to pad the beginning so all
# the columns in the final ColumnDataSource are the same length
pix, piy = makePatches(r.index, y1lim, first=True)
# The "master" data source to be used for plotting.
# Generate it via the column names in the now-merged 'r' DataFrame
# Start with the 'index' 'pix' and 'piy' since they're always those names
mds = dict(index=r.index, pix=pix, piy=piy)
# Start our plot source
cds = ColumnDataSource(mds)
# Now loop over the rest of our columns to fill it out, plotting as we go
cols = r.columns
lineSet = []
legendItems = []
for i, col in enumerate(cols):
# Add our data to the cds
cds.add(getattr(r, col), name=col)
# Make the actual line plot object
# TODO: Make this search in a given "y2" axis list
if col.lower() == "humidity":
lineObj, _ = plotLineWithPoints(p, cds, col, dset[i], yrname="y2")
else:
lineObj, _ = plotLineWithPoints(p, cds, col, dset[i])
lineSet.append(lineObj)
# Now make it's corresponding legend item
legendObj = LegendItem(label=col, renderers=[lineObj])
legendItems.append(legendObj)
legend = Legend(items=legendItems,
location="bottom_center",
orientation='horizontal',
spacing=15)
p.add_layout(legend, 'below')
# Customize the active tools
p.toolbar.autohide = True
# HACK HACK HACK HACK HACK
# Apply the patches to carry the tooltips
#
# Shouldn't I just stream this instead of pix/nix and piy/niy ???
#
simg = p.patches('pix',
'piy',
source=cds,
fill_color=None,
fill_alpha=0.0,
line_color=None)
# This will also create the tooltips for each of the entries in cols
ht = createHoverTool(simg, cols)
p.add_tools(ht)
return p, cds, cols