Plotly: How to customize the symbols of a 3D scatter plot?

In the example below:

import plotly.express as px
df = px.data.iris()
fig = px.scatter_3d(df, x='sepal_length', y='sepal_width', z='petal_width',
                    color='petal_length', symbol='species')
fig.show()

symbol is determined by the the 'species' represented by circles, rhombus and squares. This may not be very clear when sample size gets large and sample points smear the plot. How can we customise the symbols, say using circle, square and cross(or other contrasting combination of shapes)?

Update:

I applied the code on another dataset and the code :

fig = px.scatter_3d(df8, x='X', y='Y', z='Z',
                    color='P', symbol='C')

# specify trace names and symbols in a dict
symbols = {'True': 'cross',
           'False':'circle-open'}

# set all symbols in fig
for i, d in enumerate(fig.data):
    fig.data[i].marker.symbol = symbols[fig.data[i].name]

fig.show()

where C has two values: True or False. However it caught error:

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-183-ea1e2ec7dd8e> in <module>
     36 # set all symbols in fig
     37 for i, d in enumerate(fig.data):
---> 38     fig.data[i].marker.symbol = symbols[fig.data[i].name]
     39 
     40 fig.show()

KeyError: 'P, True'

Update:

Reproducible example:

{'ID': {0: '672590',
  1: '672120',
  2: '672090',
  3: '672349',
  4: '672453',
  5: '672560',
  6: '672051',
  7: '880505',
  8: '672593',
  9: '880097',
  10: '891458',
  11: '672091',
  12: '672569',
  13: '672603',
  14: '790030',
  15: '672350',
  16: '673480',
  17: 'I00042',
  18: '880297',
  19: '894620'},
 'X': {0: 0.20111215435497176,
  1: 0.21248998904335528,
  2: 0.2086689759935364,
  3: 0.22337836085443835,
  4: 0.17847099434376115,
  5: 0.24827331723865761,
  6: 0.14411891907440183,
  7: 0.20863940038267367,
  8: 0.166299824101773,
  9: 0.20548401328860527,
  10: 0.18007828100726822,
  11: 0.21887731187605308,
  12: 0.1971207940494219,
  13: 0.19247420041228508,
  14: 0.21605657330040987,
  15: 0.15779241902165092,
  16: 0.22536060645732897,
  17: 0.19268784843224268,
  18: 0.2400112771421119,
  19: 0.22548124117213691},
 'Y': {0: 2473.923076923077,
  1: 2031.1538461538462,
  2: 2383.1923076923076,
  3: 1830.7692307692307,
  4: 1780.2307692307693,
  5: 1194.8461538461538,
  6: 1641.0,
  7: 1563.3076923076924,
  8: 1246.2307692307693,
  9: 931.6153846153846,
  10: 1207.076923076923,
  11: 799.6538461538462,
  12: 560.8461538461538,
  13: 1158.076923076923,
  14: 1221.6923076923076,
  15: 3030.076923076923,
  16: 1178.076923076923,
  17: 552.3846153846154,
  18: 1380.3076923076924,
  19: 1027.5384615384614},
 'Z': {0: 385.84615384615387,
  1: 288.46153846153845,
  2: 281.9230769230769,
  3: 273.61538461538464,
  4: 252.0,
  5: 231.69230769230768,
  6: 213.30769230769232,
  7: 203.3846153846154,
  8: 191.07692307692307,
  9: 189.46153846153845,
  10: 181.07692307692307,
  11: 176.76923076923077,
  12: 173.30769230769232,
  13: 169.6153846153846,
  14: 166.15384615384616,
  15: 165.30769230769232,
  16: 160.53846153846155,
  17: 159.84615384615384,
  18: 159.0,
  19: 145.3846153846154},
 'C': {0: True,
  1: True,
  2: True,
  3: True,
  4: True,
  5: True,
  6: True,
  7: True,
  8: True,
  9: True,
  10: True,
  11: False,
  12: False,
  13: True,
  14: True,
  15: True,
  16: True,
  17: False,
  18: True,
  19: True},
 'P': {0: 'P',
  1: 'P',
  2: 'P',
  3: 'P',
  4: 'P',
  5: 'X',
  6: 'P',
  7: 'P',
  8: 'P',
  9: 'P',
  10: 'P',
  11: 'P',
  12: 'P',
  13: 'P',
  14: 'P',
  15: 'P',
  16: 'P',
  17: 'X',
  18: 'P',
  19: 'P'}}

Solution

The answer:

# specify trace names and symbols in a dict
symbols = {'setosa': 'cross',
           'versicolor':'circle-open',
           'virginica':'diamond-open'}

# set all symbols in fig
for i, d in enumerate(fig.data):
    fig.data[i].marker.symbol = symbols[fig.data[i].name]

The details:

In this case you can set any marker symbol manually using, for example:

fig.data[<i>].marker.symbol = <symbol>

Where <i> is an integer index specifying which trace you'd like to change, and the <symbol> property for a 3D scatter object is an enumeration that may be specified as one of the following enumeration values:

['circle', 'circle-open', 'square', 'square-open',
'diamond', 'diamond-open', 'cross', 'x']

Example 1 - Single trace:

fig.data[2].marker.symbol = 'circle-open'
fig.show()

Plot 1:

Example 2 - Multiple traces:

In order to make things a bit more dynamic, you can retrieve all trace names using:

['setosa', 'versicolor', 'virginica']

Then you can specify your own dictionary of names and symbols, and set all symbols for all traces using:

# specify trace names and symbols in a dict
symbols = {'setosa': 'cross',
           'versicolor':'circle-open',
           'virginica':'diamond-open'}

# set all symbols in fig
for i, d in enumerate(fig.data):
    fig.data[i].marker.symbol = symbols[fig.data[i].name]

fig.show()

Plot 2:

Complete code:

import plotly.express as px
df = px.data.iris()
fig = px.scatter_3d(df, x='sepal_length', y='sepal_width', z='petal_width',
                    color='petal_length', symbol='species')
#fig.show()

# specify trace names and symbols in a dict
symbols = {'setosa': 'cross',
           'versicolor':'circle-open',
           'virginica':'diamond-open'}

# set all symbols in fig
for i, d in enumerate(fig.data):
    fig.data[i].marker.symbol = symbols[fig.data[i].name]

fig.show()

Edit: Answer to addendum by OP

What happens here is that you're assigning colors and symbols in px.scatter3d using color='P', symbol='C'. And that has consequences for the trace names which are now, for example, 'name': 'P, True'. This leads the following to break:

for i, d in enumerate(fig.data):
    fig.data[i].marker.symbol = symbols[fig.data[i].name]

Judging by your defined symbols dict:

# specify trace names and symbols in a dict
symbols = {'True': 'cross',
           'False':'circle-open'}

It would appear that you'd only like to distinguish your symbols by the last part of 'name': 'P, True' whics is True or False. And you can subset your symbols dict by specifying that in symbols[fig.data[i].name.split(', ')[1]] and thus get:

Complete code with data sample:

import pandas as pd
import plotly.express as px

df8 = pd.DataFrame({'ID': {0: '672590',
  1: '672120',
  2: '672090',
  3: '672349',
  4: '672453',
  5: '672560',
  6: '672051',
  7: '880505',
  8: '672593',
  9: '880097',
  10: '891458',
  11: '672091',
  12: '672569',
  13: '672603',
  14: '790030',
  15: '672350',
  16: '673480',
  17: 'I00042',
  18: '880297',
  19: '894620'},
 'X': {0: 0.20111215435497176,
  1: 0.21248998904335528,
  2: 0.2086689759935364,
  3: 0.22337836085443835,
  4: 0.17847099434376115,
  5: 0.24827331723865761,
  6: 0.14411891907440183,
  7: 0.20863940038267367,
  8: 0.166299824101773,
  9: 0.20548401328860527,
  10: 0.18007828100726822,
  11: 0.21887731187605308,
  12: 0.1971207940494219,
  13: 0.19247420041228508,
  14: 0.21605657330040987,
  15: 0.15779241902165092,
  16: 0.22536060645732897,
  17: 0.19268784843224268,
  18: 0.2400112771421119,
  19: 0.22548124117213691},
 'Y': {0: 2473.923076923077,
  1: 2031.1538461538462,
  2: 2383.1923076923076,
  3: 1830.7692307692307,
  4: 1780.2307692307693,
  5: 1194.8461538461538,
  6: 1641.0,
  7: 1563.3076923076924,
  8: 1246.2307692307693,
  9: 931.6153846153846,
  10: 1207.076923076923,
  11: 799.6538461538462,
  12: 560.8461538461538,
  13: 1158.076923076923,
  14: 1221.6923076923076,
  15: 3030.076923076923,
  16: 1178.076923076923,
  17: 552.3846153846154,
  18: 1380.3076923076924,
  19: 1027.5384615384614},
 'Z': {0: 385.84615384615387,
  1: 288.46153846153845,
  2: 281.9230769230769,
  3: 273.61538461538464,
  4: 252.0,
  5: 231.69230769230768,
  6: 213.30769230769232,
  7: 203.3846153846154,
  8: 191.07692307692307,
  9: 189.46153846153845,
  10: 181.07692307692307,
  11: 176.76923076923077,
  12: 173.30769230769232,
  13: 169.6153846153846,
  14: 166.15384615384616,
  15: 165.30769230769232,
  16: 160.53846153846155,
  17: 159.84615384615384,
  18: 159.0,
  19: 145.3846153846154},
 'C': {0: True,
  1: True,
  2: True,
  3: True,
  4: True,
  5: True,
  6: True,
  7: True,
  8: True,
  9: True,
  10: True,
  11: False,
  12: False,
  13: True,
  14: True,
  15: True,
  16: True,
  17: False,
  18: True,
  19: True},
 'P': {0: 'P',
  1: 'P',
  2: 'P',
  3: 'P',
  4: 'P',
  5: 'X',
  6: 'P',
  7: 'P',
  8: 'P',
  9: 'P',
  10: 'P',
  11: 'P',
  12: 'P',
  13: 'P',
  14: 'P',
  15: 'P',
  16: 'P',
  17: 'X',
  18: 'P',
  19: 'P'}})

fig = px.scatter_3d(df8, x='X', y='Y', z='Z',
                    color='P', symbol='C')

# specify trace names and symbols in a dict
symbols = {'True': 'cross',
           'False':'circle-open'}

# set all symbols in fig
for i, d in enumerate(fig.data):
    fig.data[i].marker.symbol = symbols[fig.data[i].name.split(', ')[1]]

fig.show()