Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Commit cf48f62

Browse files
author
Kevin Yan
committed
Merge pull request plotly#473 from plotly/distplot-fix
Distplot fix
2 parents dfef7af + bda2531 commit cf48f62

File tree

2 files changed

+194
-23
lines changed

2 files changed

+194
-23
lines changed

‎plotly/tests/test_optional/test_figure_factory.py

Lines changed: 175 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -44,13 +44,60 @@ def test_unequal_data_label_length(self):
4444
self.assertRaises(PlotlyError, tls.FigureFactory.create_distplot,
4545
**kwargs)
4646

47-
def test_simple_distplot(self):
47+
def test_simple_distplot_prob_density(self):
4848

4949
# we should be able to create a single distplot with a simple dataset
5050
# and default kwargs
5151

5252
dp = tls.FigureFactory.create_distplot(hist_data=[[1, 2, 2, 3]],
53-
group_labels=['distplot'])
53+
group_labels=['distplot'],
54+
histnorm='probability density')
55+
expected_dp_layout = {'barmode': 'overlay',
56+
'hovermode': 'closest',
57+
'legend': {'traceorder': 'reversed'},
58+
'xaxis1': {'anchor': 'y2', 'domain': [0.0, 1.0], 'zeroline': False},
59+
'yaxis1': {'anchor': 'free', 'domain': [0.35, 1], 'position': 0.0},
60+
'yaxis2': {'anchor': 'x1',
61+
'domain': [0, 0.25],
62+
'dtick': 1,
63+
'showticklabels': False}}
64+
self.assertEqual(dp['layout'], expected_dp_layout)
65+
66+
expected_dp_data_hist = {'autobinx': False,
67+
'histnorm': 'probability density',
68+
'legendgroup': 'distplot',
69+
'marker': {'color': 'rgb(31, 119, 180)'},
70+
'name': 'distplot',
71+
'opacity': 0.7,
72+
'type': 'histogram',
73+
'x': [1, 2, 2, 3],
74+
'xaxis': 'x1',
75+
'xbins': {'end': 3.0, 'size': 1.0, 'start': 1.0},
76+
'yaxis': 'y1'}
77+
self.assertEqual(dp['data'][0], expected_dp_data_hist)
78+
79+
expected_dp_data_rug = {'legendgroup': 'distplot',
80+
'marker': {'color': 'rgb(31, 119, 180)',
81+
'symbol': 'line-ns-open'},
82+
'mode': 'markers',
83+
'name': 'distplot',
84+
'showlegend': False,
85+
'text': None,
86+
'type': 'scatter',
87+
'x': [1, 2, 2, 3],
88+
'xaxis': 'x1',
89+
'y': ['distplot', 'distplot',
90+
'distplot', 'distplot'],
91+
'yaxis': 'y2'}
92+
self.assertEqual(dp['data'][2], expected_dp_data_rug)
93+
94+
def test_simple_distplot_prob(self):
95+
96+
# we should be able to create a single distplot with a simple dataset
97+
# and default kwargs
98+
99+
dp = tls.FigureFactory.create_distplot(hist_data=[[1, 2, 2, 3]],
100+
group_labels=['distplot'], histnorm='probability')
54101
expected_dp_layout = {'barmode': 'overlay',
55102
'hovermode': 'closest',
56103
'legend': {'traceorder': 'reversed'},
@@ -90,7 +137,7 @@ def test_simple_distplot(self):
90137
'yaxis': 'y2'}
91138
self.assertEqual(dp['data'][2], expected_dp_data_rug)
92139

93-
def test_distplot_more_args(self):
140+
def test_distplot_more_args_prob_dens(self):
94141

95142
# we should be able to create a distplot with 2 datasets no
96143
# rugplot, defined bin_size, and added title
@@ -106,6 +153,69 @@ def test_distplot_more_args(self):
106153
group_labels = ['2012', '2013']
107154

108155
dp = tls.FigureFactory.create_distplot(hist_data, group_labels,
156+
histnorm='probability density',
157+
show_rug=False, bin_size=.2)
158+
dp['layout'].update(title='Dist Plot')
159+
160+
expected_dp_layout = {'barmode': 'overlay',
161+
'hovermode': 'closest',
162+
'legend': {'traceorder': 'reversed'},
163+
'title': 'Dist Plot',
164+
'xaxis1': {'anchor': 'y2', 'domain': [0.0, 1.0],
165+
'zeroline': False},
166+
'yaxis1': {'anchor': 'free', 'domain': [0.0, 1],
167+
'position': 0.0}}
168+
self.assertEqual(dp['layout'], expected_dp_layout)
169+
170+
expected_dp_data_hist_1 = {'autobinx': False,
171+
'histnorm': 'probability density',
172+
'legendgroup': '2012',
173+
'marker': {'color': 'rgb(31, 119, 180)'},
174+
'name': '2012',
175+
'opacity': 0.7,
176+
'type': 'histogram',
177+
'x': [0.8, 1.2, 0.2, 0.6, 1.6, -0.9, -0.07,
178+
1.95, 0.9, -0.2, -0.5, 0.3, 0.4,
179+
-0.37, 0.6],
180+
'xaxis': 'x1',
181+
'xbins': {'end': 1.95, 'size': 0.2,
182+
'start': -0.9},
183+
'yaxis': 'y1'}
184+
self.assertEqual(dp['data'][0], expected_dp_data_hist_1)
185+
186+
expected_dp_data_hist_2 = {'autobinx': False,
187+
'histnorm': 'probability density',
188+
'legendgroup': '2013',
189+
'marker': {'color': 'rgb(255, 127, 14)'},
190+
'name': '2013',
191+
'opacity': 0.7,
192+
'type': 'histogram',
193+
'x': [0.8, 1.5, 1.5, 0.6, 0.59, 1.0, 0.8,
194+
1.7, 0.5, 0.8, -0.3, 1.2, 0.56, 0.3,
195+
2.2],
196+
'xaxis': 'x1',
197+
'xbins': {'end': 2.2, 'size': 0.2,
198+
'start': -0.3},
199+
'yaxis': 'y1'}
200+
self.assertEqual(dp['data'][1], expected_dp_data_hist_2)
201+
202+
def test_distplot_more_args_prob(self):
203+
204+
# we should be able to create a distplot with 2 datasets no
205+
# rugplot, defined bin_size, and added title
206+
207+
hist1_x = [0.8, 1.2, 0.2, 0.6, 1.6,
208+
-0.9, -0.07, 1.95, 0.9, -0.2,
209+
-0.5, 0.3, 0.4, -0.37, 0.6]
210+
hist2_x = [0.8, 1.5, 1.5, 0.6, 0.59,
211+
1.0, 0.8, 1.7, 0.5, 0.8,
212+
-0.3, 1.2, 0.56, 0.3, 2.2]
213+
214+
hist_data = [hist1_x] + [hist2_x]
215+
group_labels = ['2012', '2013']
216+
217+
dp = tls.FigureFactory.create_distplot(hist_data, group_labels,
218+
histnorm='probability',
109219
show_rug=False, bin_size=.2)
110220
dp['layout'].update(title='Dist Plot')
111221

@@ -151,7 +261,55 @@ def test_distplot_more_args(self):
151261
'yaxis': 'y1'}
152262
self.assertEqual(dp['data'][1], expected_dp_data_hist_2)
153263

154-
def test_distplot_binsize_array(self):
264+
def test_distplot_binsize_array_prob(self):
265+
hist1_x = [0.8, 1.2, 0.2, 0.6, 1.6, -0.9, -0.07, 1.95, 0.9, -0.2,
266+
-0.5, 0.3, 0.4, -0.37, 0.6]
267+
hist2_x = [0.8, 1.5, 1.5, 0.6, 0.59, 1.0, 0.8, 1.7, 0.5, 0.8, -0.3,
268+
1.2, 0.56, 0.3, 2.2]
269+
270+
hist_data = [hist1_x, hist2_x]
271+
group_labels = ['2012', '2013']
272+
273+
dp = tls.FigureFactory.create_distplot(hist_data, group_labels,
274+
histnorm='probability',
275+
show_rug=False,
276+
bin_size=[.2, .2])
277+
278+
expected_dp_data_hist_1 = {'autobinx': False,
279+
'histnorm': 'probability density',
280+
'legendgroup': '2012',
281+
'marker':
282+
{'color': 'rgb(31, 119, 180)'},
283+
'name': '2012',
284+
'opacity': 0.7,
285+
'type': 'histogram',
286+
'x': [0.8, 1.2, 0.2, 0.6, 1.6, -0.9,
287+
-0.07, 1.95, 0.9, -0.2, -0.5, 0.3,
288+
0.4, -0.37, 0.6],
289+
'xaxis': 'x1',
290+
'xbins': {'end': 1.95, 'size': 0.2,
291+
'start': -0.9},
292+
'yaxis': 'y1'}
293+
self.assertEqual(dp['data'][0], expected_dp_data_hist_1)
294+
295+
expected_dp_data_hist_2 = {'autobinx': False,
296+
'histnorm': 'probability density',
297+
'legendgroup': '2013',
298+
'marker':
299+
{'color': 'rgb(255, 127, 14)'},
300+
'name': '2013',
301+
'opacity': 0.7,
302+
'type': 'histogram',
303+
'x': [0.8, 1.5, 1.5, 0.6, 0.59, 1.0,
304+
0.8, 1.7, 0.5, 0.8, -0.3, 1.2,
305+
0.56, 0.3, 2.2],
306+
'xaxis': 'x1',
307+
'xbins': {'end': 2.2, 'size': 0.2,
308+
'start': -0.3},
309+
'yaxis': 'y1'}
310+
self.assertEqual(dp['data'][1], expected_dp_data_hist_2)
311+
312+
def test_distplot_binsize_array_prob_density(self):
155313
hist1_x = [0.8, 1.2, 0.2, 0.6, 1.6, -0.9, -0.07, 1.95, 0.9, -0.2,
156314
-0.5, 0.3, 0.4, -0.37, 0.6]
157315
hist2_x = [0.8, 1.5, 1.5, 0.6, 0.59, 1.0, 0.8, 1.7, 0.5, 0.8, -0.3,
@@ -161,35 +319,38 @@ def test_distplot_binsize_array(self):
161319
group_labels = ['2012', '2013']
162320

163321
dp = tls.FigureFactory.create_distplot(hist_data, group_labels,
322+
histnorm='probability',
164323
show_rug=False,
165324
bin_size=[.2, .2])
166325

167326
expected_dp_data_hist_1 = {'autobinx': False,
168-
'histnorm': 'probability',
327+
'histnorm': 'probability density',
169328
'legendgroup': '2012',
170-
'marker': {'color': 'rgb(31, 119, 180)'},
329+
'marker':
330+
{'color': 'rgb(31, 119, 180)'},
171331
'name': '2012',
172332
'opacity': 0.7,
173333
'type': 'histogram',
174-
'x': [0.8, 1.2, 0.2, 0.6, 1.6, -0.9, -0.07,
175-
1.95, 0.9, -0.2, -0.5, 0.3, 0.4,
176-
-0.37, 0.6],
334+
'x': [0.8, 1.2, 0.2, 0.6, 1.6, -0.9,
335+
-0.07, 1.95, 0.9, -0.2, -0.5, 0.3,
336+
0.4, -0.37, 0.6],
177337
'xaxis': 'x1',
178338
'xbins': {'end': 1.95, 'size': 0.2,
179339
'start': -0.9},
180340
'yaxis': 'y1'}
181341
self.assertEqual(dp['data'][0], expected_dp_data_hist_1)
182342

183343
expected_dp_data_hist_2 = {'autobinx': False,
184-
'histnorm': 'probability',
344+
'histnorm': 'probability density',
185345
'legendgroup': '2013',
186-
'marker': {'color': 'rgb(255, 127, 14)'},
346+
'marker':
347+
{'color': 'rgb(255, 127, 14)'},
187348
'name': '2013',
188349
'opacity': 0.7,
189350
'type': 'histogram',
190-
'x': [0.8, 1.5, 1.5, 0.6, 0.59, 1.0, 0.8,
191-
1.7, 0.5, 0.8, -0.3, 1.2, 0.56, 0.3,
192-
2.2],
351+
'x': [0.8, 1.5, 1.5, 0.6, 0.59, 1.0,
352+
0.8, 1.7, 0.5, 0.8, -0.3, 1.2,
353+
0.56, 0.3, 2.2],
193354
'xaxis': 'x1',
194355
'xbins': {'end': 2.2, 'size': 0.2,
195356
'start': -0.3},

‎plotly/tools.py

Lines changed: 19 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -28,6 +28,9 @@
2828
'rgb(227, 119, 194)', 'rgb(127, 127, 127)',
2929
'rgb(188, 189, 34)', 'rgb(23, 190, 207)']
3030

31+
DEFAULT_HISTNORM = 'probability density'
32+
ALTERNATIVE_HISTNORM = 'probability'
33+
3134

3235
# Warning format
3336
def warning_on_one_line(message, category, filename, lineno,
@@ -4245,7 +4248,7 @@ def create_candlestick(open, high, low, close,
42454248
@staticmethod
42464249
def create_distplot(hist_data, group_labels,
42474250
bin_size=1., curve_type='kde',
4248-
colors=[], rug_text=[],
4251+
colors=[], rug_text=[], histnorm=DEFAULT_HISTNORM,
42494252
show_hist=True, show_curve=True,
42504253
show_rug=True):
42514254
"""
@@ -4262,6 +4265,8 @@ def create_distplot(hist_data, group_labels,
42624265
:param (list[float]|float) bin_size: Size of histogram bins.
42634266
Default = 1.
42644267
:param (str) curve_type: 'kde' or 'normal'. Default = 'kde'
4268+
:param (str) histnorm: 'probability density' or 'probability'
4269+
Default = 'probability density'
42654270
:param (bool) show_hist: Add histogram to distplot? Default = True
42664271
:param (bool) show_curve: Add curve to distplot? Default = True
42674272
:param (bool) show_rug: Add rug to distplot? Default = True
@@ -4370,23 +4375,23 @@ def create_distplot(hist_data, group_labels,
43704375
bin_size = [bin_size]*len(hist_data)
43714376

43724377
hist = _Distplot(
4373-
hist_data, group_labels, bin_size,
4378+
hist_data, histnorm, group_labels, bin_size,
43744379
curve_type, colors, rug_text,
43754380
show_hist, show_curve).make_hist()
43764381

43774382
if curve_type == 'normal':
43784383
curve = _Distplot(
4379-
hist_data, group_labels, bin_size,
4384+
hist_data, histnorm, group_labels, bin_size,
43804385
curve_type, colors, rug_text,
43814386
show_hist, show_curve).make_normal()
43824387
else:
43834388
curve = _Distplot(
4384-
hist_data, group_labels, bin_size,
4389+
hist_data, histnorm, group_labels, bin_size,
43854390
curve_type, colors, rug_text,
43864391
show_hist, show_curve).make_kde()
43874392

43884393
rug = _Distplot(
4389-
hist_data, group_labels, bin_size,
4394+
hist_data, histnorm, group_labels, bin_size,
43904395
curve_type, colors, rug_text,
43914396
show_hist, show_curve).make_rug()
43924397

@@ -5262,10 +5267,11 @@ class _Distplot(FigureFactory):
52625267
"""
52635268
Refer to TraceFactory.create_distplot() for docstring
52645269
"""
5265-
def __init__(self, hist_data, group_labels,
5270+
def __init__(self, hist_data, histnorm, group_labels,
52665271
bin_size, curve_type, colors,
52675272
rug_text, show_hist, show_curve):
52685273
self.hist_data = hist_data
5274+
self.histnorm = histnorm
52695275
self.group_labels = group_labels
52705276
self.bin_size = bin_size
52715277
self.show_hist = show_hist
@@ -5307,7 +5313,7 @@ def make_hist(self):
53075313
x=self.hist_data[index],
53085314
xaxis='x1',
53095315
yaxis='y1',
5310-
histnorm='probability',
5316+
histnorm=self.histnorm,
53115317
name=self.group_labels[index],
53125318
legendgroup=self.group_labels[index],
53135319
marker=dict(color=self.colors[index]),
@@ -5334,7 +5340,9 @@ def make_kde(self):
53345340
self.curve_y[index] = (scipy.stats.gaussian_kde
53355341
(self.hist_data[index])
53365342
(self.curve_x[index]))
5337-
self.curve_y[index] *= self.bin_size[index]
5343+
5344+
if self.histnorm == ALTERNATIVE_HISTNORM:
5345+
self.curve_y[index] *= self.bin_size[index]
53385346

53395347
for index in range(self.trace_number):
53405348
curve[index] = dict(type='scatter',
@@ -5369,7 +5377,9 @@ def make_normal(self):
53695377
/ 500 for x in range(500)]
53705378
self.curve_y[index] = scipy.stats.norm.pdf(
53715379
self.curve_x[index], loc=mean[index], scale=sd[index])
5372-
self.curve_y[index] *= self.bin_size[index]
5380+
5381+
if self.histnorm == ALTERNATIVE_HISTNORM:
5382+
self.curve_y[index] *= self.bin_size[index]
53735383

53745384
for index in range(self.trace_number):
53755385
curve[index] = dict(type='scatter',

0 commit comments

Comments
 (0)