如何更新我的Bokeh Legend以反映Pandas数据帧中的类别变量

Question

我正试图用Bokeh制作一个下拉菜单，突出显示我找到的聚类中的点。我有下拉菜单，但现在我想能够通过颜色可视化另一个分类变量：具有Masc、Fem和Neuter级别的Noun Class。问题是，当我切换正在可视化的集群时，图例不会更新。此外，如果我看到的第一个集群中没有所有3个名词类，代码就会开始将我试图查看的所有其他集群视为(错误地(具有第一个集群的名词类。例如，如果集群0是默认的，并且只有Masc点，那么我使用下拉菜单查看的所有其他集群都被视为只有Mask点，即使它们在实际DF中有Fem或Neuter。

我的主要问题是：我如何更新图例，使其只关注"Curr"的相应名词类

这里有一些可复制的代码：

import pandas as pd
from bokeh.io import output_file, show, output_notebook, save, push_notebook
from bokeh.models import ColumnDataSource, Select, DateRangeSlider, CustomJS
from bokeh.plotting import figure, Figure, show
from bokeh.models import CustomJS 
from bokeh.layouts import row,column,layout
import random
import numpy as np
from bokeh.transform import factor_cmap
from bokeh.palettes import Colorblind
import bokeh.io
from bokeh.resources import INLINE
#Generate reproducible DF
noun_class_names = ["Masc","Fem","Neuter"]
x = [random.randint(0,50) for i in range(100)]
y = [random.randint(0,50) for i in range(100)]
rand_clusters = [str(random.randint(0,10)) for i in range(100)]
noun_classes = [random.choice(noun_class_names) for i in range(100)]
df = pd.DataFrame({'x_coord':x, 'y_coord':y,'noun class':noun_classes,'cluster labels':rand_clusters})
df.loc[df['cluster labels'] == '0', 'noun class'] = 'Masc' #ensure that cluster 0 has all same noun class to illustrate error

clusters = [str(i) for i in range(len(df['cluster labels'].unique()))]
cols1 = df#[['cluster labels','x_coord', 'y_coord']]
cols2 = cols1[cols1['cluster labels'] == '0']
Overall = ColumnDataSource(data=cols1)
Curr = ColumnDataSource(data=cols2)

#plot and the menu is linked with each other by this callback function
callback = CustomJS(args=dict(source=Overall, sc=Curr), code="""
var f = cb_obj.value
sc.data['x_coord']=[]
sc.data['y_coord']=[]
for(var i = 0; i <= source.get_length(); i++){
if (source.data['cluster labels'][i] == f){
sc.data['x_coord'].push(source.data['x_coord'][i])
sc.data['y_coord'].push(source.data['y_coord'][i])
sc.data['noun class'].push(source.data['noun class'][i])
sc.data['cluster labels'].push(source.data['cluster labels'][i])
}
}   
sc.change.emit();
""")
menu = Select(options=clusters, value='0', title = 'Cluster #')  # create drop down menu
bokeh_p=figure(x_axis_label ='X Coord', y_axis_label = 'Y Coord', y_axis_type="linear",x_axis_type="linear") #creating figure object 
mapper = factor_cmap(field_name = "noun class", palette = Colorblind[6], factors = df['noun class'].unique()) #color mapper for noun classes
bokeh_p.circle(x='x_coord', y='y_coord', color='gray', alpha = .5, source=Overall) #plot all other points in gray
bokeh_p.circle(x='x_coord', y='y_coord', color=mapper, line_width = 1, source=Curr, legend_group = 'noun class') # plotting the desired cluster using glyph circle and colormapper
bokeh_p.legend.title = "Noun Classes"

menu.js_on_change('value', callback) # calling the function on change of selection
bokeh.io.output_notebook(INLINE)
show(layout(menu,bokeh_p), notebook_handle=True)

提前谢谢，我希望你今天过得愉快：(

Answer 1

伊玛和你们一起保持真实。。。代码按照我现在想要的方式工作，我不完全确定我做了什么。我想我所做的是重置Curr数据源中的名词类，然后在选择新集群以可视化和更新xy坐标后更新图例标签字段。如果有人能为了子孙后代的利益确认或纠正我，我将不胜感激：(

最好！

import pandas as pd
import random
import numpy as np
from bokeh.plotting import figure, Figure, show
from bokeh.io import output_notebook, push_notebook, show, output_file, save
from bokeh.transform import factor_cmap
from bokeh.palettes import Colorblind
from bokeh.layouts import layout, gridplot, column, row
from bokeh.models import ColumnDataSource, Slider, CustomJS, Select, DateRangeSlider, Legend, LegendItem
import bokeh.io
from bokeh.resources import INLINE
#Generate reproducible DF
noun_class_names = ["Masc","Fem","Neuter"]
x = [random.randint(0,50) for i in range(100)]
y = [random.randint(0,50) for i in range(100)]
rand_clusters = [str(random.randint(0,10)) for i in range(100)]
noun_classes = [random.choice(noun_class_names) for i in range(100)]
df = pd.DataFrame({'x_coord':x, 'y_coord':y,'noun class':noun_classes,'cluster labels':rand_clusters})
df.loc[df['cluster labels'] == '0', 'noun class'] = 'Masc' #ensure that cluster 0 has all same noun class to illustrate error
clusters = [str(i) for i in range(len(df['cluster labels'].unique()))]
cols1 = df#[['cluster labels','x_coord', 'y_coord']]
cols2 = cols1[cols1['cluster labels'] == '0']
Overall = ColumnDataSource(data=cols1)
Curr = ColumnDataSource(data=cols2)

#plot and the menu is linked with each other by this callback function
callback = CustomJS(args=dict(source=Overall, sc=Curr), code="""
var f = cb_obj.value
sc.data['x_coord']=[]
sc.data['y_coord']=[]
sc.data['noun class'] =[]
for(var i = 0; i <= source.get_length(); i++){
if (source.data['cluster labels'][i] == f){
sc.data['x_coord'].push(source.data['x_coord'][i])
sc.data['y_coord'].push(source.data['y_coord'][i])
sc.data['noun class'].push(source.data['noun class'][i])
sc.data['cluster labels'].push(source.data['cluster labels'][i])
}
}
sc.change.emit();
bokeh_p.legend.label.field = sc.data['noun class'];

""")
menu = Select(options=clusters, value='0', title = 'Cluster #')  # create drop down menu
bokeh_p=figure(x_axis_label ='X Coord', y_axis_label = 'Y Coord', y_axis_type="linear",x_axis_type="linear") #creating figure object 
mapper = factor_cmap(field_name = "noun class", palette = Colorblind[6], factors = df['noun class'].unique()) #color mapper- sorry this was a thing that carried over from og code (fixed now)
bokeh_p.circle(x='x_coord', y='y_coord', color='gray', alpha = .05, source=Overall)
bokeh_p.circle(x = 'x_coord', y = 'y_coord', fill_color = mapper, line_color = mapper, source = Curr, legend_field = 'noun class')

bokeh_p.legend.title = "Noun Classes"

menu.js_on_change('value', callback) # calling the function on change of selection
bokeh.io.output_notebook(INLINE)
show(layout(menu,bokeh_p), notebook_handle=True)