虚化。在嵌套的分类图上加一条线。

Question

我有一个嵌套分类x轴的Bokeh图。这是一个玩具问题的代码。我的真实用例自然是一个更大、更复杂的数据集。

import pandas as pd
from bokeh.io import output_notebook, show, reset_output
from bokeh.models import Band, Span, FactorRange, ColumnDataSource
from bokeh.plotting import figure

reset_output()
output_notebook()

data = {'fruit': ['Apples', 'Pears'],
        '2015': [2, 1],
        '2016': [5, 3]}

tidy_df = (pd.DataFrame(data)
           .melt(id_vars=["fruit"], var_name="year")
           .assign(fruit_year=lambda df: list(zip(df['fruit'], df['year'])))
           .set_index('fruit_year'))

display(tidy_df)

p = figure(x_range=FactorRange(factors=tidy_df.index.unique()),
           height=300,
           width=300)

cds = ColumnDataSource(tidy_df)

p.circle(x='fruit_year', 
         y='value', 
         size=20,
         source=cds,
         line_color=None,
        )

# this does not show anything or cause an error
p.line(
    x=[("Apples", 2015), ("Apples", 2016)], 
    y=[3.5, 3.5], 
    color="red", 
    line_width=2
)

# this works, but does not scale to problems where location can't be manually specified
# also, the line does not line up with the data?
p.line(
    x=[4, 5], 
    y=[2, 2], 
    color="red", 
    line_width=2
)


show(p)

输出。

这条线没有出现在图上，也没有出错。

p.line(
    x=[("Apples", 2015), ("Apples", 2016)], 
    y=[3.5, 3.5], 
    color="red", 
    line_width=2
)

我如何指定 x 来让这一行显示出来？我是否可以指定一个任意的子级别，即只针对 ("Apples", 2015)?

Answer 1

非常类似的问题解决方案 此处. (我提出了这个问题，并在这个答案的基础上建立起来。)

一般的概念是基于一个人的初始数据集创建数据框，然后建立多个 ColumnDataSource的数据帧。

下面是完整的代码。

p = figure(x_range=FactorRange(factors=tidy_df.index.unique()),
           plot_height=400,
           plot_width=400,
           tooltips=[('Fruit', '@fruit'), # first string is user-defined; second string must refer to a column
                     ('Year', '@year'),
                     ('Value', '@value')])

cds = ColumnDataSource(tidy_df)

index_cmap = factor_cmap("fruit", 
                         Spectral5[:2], 
                         factors=sorted(tidy_df["fruit"].unique())) # this is a reference back to the dataframe

p.circle(x='fruit_year', 
         y='value', 
         size=20,
         source=cds,
         fill_color=index_cmap,
         line_color=None,
        )

# add global median
# how to add for each fruit?
median = Span(location=tidy_df["value"].median(), # median value for Apples
              #dimension='height', 
              line_color='orange',
              line_dash='dashed', 
              line_width=1.0
             )

p.add_layout(median)


for fruit, stddev in list(zip(tidy_df["fruit"].unique(), tidy_df.groupby("fruit").std().values.flatten())):
    b_df = tidy_df[tidy_df['fruit'] == fruit]\
        .drop(columns=['fruit', 'year'])\
        .assign(lower=lambda df: df['value'].median() - stddev,
                upper=lambda df: df['value'].median() + stddev)\
        .assign(median=lambda df: df["value"].median())\
        .drop(columns='value')
    display(b_df)

    # create another cds
    cds2 = ColumnDataSource(b_df)

    p.add_layout(
        Band(
            base='fruit_year', 
            lower='lower', 
            upper='upper',
            source=cds2)
    )

    p.line(x="fruit_year",
           y="median",
           source=cds2,
           color="red",
           line_width=2, 
           line_dash='dashed', 
          )

show(p)