在python中使用bokeh创建堆积的条形图

Question

I have the below data which captures the current class of a device and whether a class change has been affected or not.

Class   Class_Change
S        yes
S        yes
G        yes
P        yes
P        yes
V        no
G        yes
V        no
V        no
V        yes
P        no

Now I want to display the Yes/No across each class in a stacked bar chart. Something like below which I have created in excel. The table is a count of yes/no across each class and the chart is the stacked bar chart of it.

I have tried the below code:

df2 = pd.DataFrame({'count': df_class.groupby(["Class","Class_Change"]).size()}).reset_index()

class = df2['Class'].tolist()
class_change = df2['Class_Change'].tolist()
count = df2['count'].tolist()

source = ColumnDataSource(data=dict(Classes = class, count=count, ClassChange = class_change,color = Viridis5))

plot = figure(x_range=tiers ,y_range=(0,max(count)), plot_height=350, plot_width = 800,title="Counts",
           toolbar_location=None, tools="")

labels = LabelSet(x = 'Class', y= 'count' , text='count', level='glyph',
                       y_offset=0 , source=source, render_mode='canvas')

plot.vbar_stack(class_change, x='class', width=0.9, color='color', source=source, legend=[value(x) for x in class_change]) 

but it is giving the error:

AttributeError: 'Figure' object has no attribute 'vbar_stack'

Can someone please help me with this??

Answer 1

The bokeh reference docs gives an example that can be reworked for your case, here :

Assuming a ColumnDataSource named source with columns 2106 and 2017, then the >following call to vbar_stack will will create two VBar renderers that stack:

p.vbar_stack(['2016', '2017'], x=10, width=0.9, color=['blue', 'red'], source=source)

vbar_stack requires that the data it is given is structured in a specific way. The values that we want stacked must be represented as columns in our column data source. Something like this:

Then we could write something like this:

plot.vbar_stack(['no', 'yes'], 
                x='Class', 
                width=0.9, 
                color=['blue', 'orange'],
                line_color='white',
                legend=['No', 'Yes'],  # caps prevents calling columns
                source=source)

However, you are running into another issue. "'Figure' object has no attribute 'vbar_stack'" tells us that the vbar_stack method is not defined for the figure object that you are trying to call this method on. The most likely cause is that you are using a version of Bokeh from before vbar_stack was added. In that case the solution is to update bokeh to the latest version with either pip or conda.