如何在Bokeh curdoc构建的交互图表中添加线性回归线

实现逻辑 #

线性回归线的计算与绘制逻辑直接嵌入现有create_figure()函数即可，完全适配当前图表重绘更新的结构，不需要修改原有控件绑定的update方法。额外增加离散轴判断逻辑：仅当X、Y轴同时选中连续型数值字段时才计算绘制回归线，避免分类字段触发计算报错。

操作步骤 #

• 补充导入numpy依赖，用于线性回归参数计算，同时替换原有代码中已被新版Pandas弃用的pd.np调用
• 在散点绘制完成后，增加轴类型判断：若X/Y均为连续字段，提取当前绘图用的x、y数值序列
• 用numpy一次多项式拟合计算线性回归的斜率、截距参数
• 取X轴取值范围的最小、最大值作为回归线端点的x坐标，代入回归公式计算对应y坐标
• 调用p.line()将回归线绘制到图表上，设置差异化样式和散点做区分

完整修改后代码 #

import pandas as pd
import numpy as np

from bokeh.layouts import column, row
from bokeh.models import Select
from bokeh.palettes import Spectral5
from bokeh.plotting import curdoc, figure
from bokeh.sampledata.autompg import autompg_clean as df

df = df.copy()

SIZES = list(range(6, 22, 3))
COLORS = Spectral5
N_SIZES = len(SIZES)
N_COLORS = len(COLORS)

# 数据清洗
df.cyl = df.cyl.astype(str)
df.yr = df.yr.astype(str)
del df['name']

columns = sorted(df.columns)
discrete = [x for x in columns if df[x].dtype == object]
continuous = [x for x in columns if x not in discrete]

def create_figure():
    xs = df[x.value].values
    ys = df[y.value].values
    x_title = x.value.title()
    y_title = y.value.title()

    kw = dict()
    if x.value in discrete:
        kw['x_range'] = sorted(set(xs))
    if y.value in discrete:
        kw['y_range'] = sorted(set(ys))
    kw['title'] = "%s vs %s" % (x_title, y_title)

    p = figure(height=600, width=800, tools='pan,box_zoom,hover,reset', **kw)
    p.xaxis.axis_label = x_title
    p.yaxis.axis_label = y_title

    if x.value in discrete:
        # 替换已弃用的pd.np调用
        p.xaxis.major_label_orientation = np.pi / 4

    sz = 9
    if size.value != 'None':
        if len(set(df[size.value])) > N_SIZES:
            groups = pd.qcut(df[size.value].values, N_SIZES, duplicates='drop')
        else:
            groups = pd.Categorical(df[size.value])
        sz = [SIZES[xx] for xx in groups.codes]

    c = "#31AADE"
    if color.value != 'None':
        if len(set(df[color.value])) > N_COLORS:
            groups = pd.qcut(df[color.value].values, N_COLORS, duplicates='drop')
        else:
            groups = pd.Categorical(df[color.value])
        c = [COLORS[xx] for xx in groups.codes]

    # 绘制原始散点
    p.circle(x=xs, y=ys, color=c, size=sz, line_color="white", alpha=0.6, hover_color='white', hover_alpha=0.5)

    # ========== 新增：线性回归线绘制逻辑 ==========
    # 仅当XY轴均为连续数值型时绘制回归线
    if x.value in continuous and y.value in continuous:
        # 转换数据类型避免计算报错
        x_num = xs.astype(np.float64)
        y_num = ys.astype(np.float64)
        # 一次多项式拟合，等价于线性回归，返回斜率、截距
        slope, intercept = np.polyfit(x_num, y_num, deg=1)
        # 取X轴两端点生成回归线坐标
        x_line = np.array([x_num.min(), x_num.max()])
        y_line = slope * x_line + intercept
        # 绘制回归线，设置差异化样式
        p.line(x=x_line, y=y_line, color="#E63946", line_width=3, line_dash="dashed", legend_label="线性回归线")
        # 配置图例样式
        p.legend.location = "top_right"
        p.legend.background_fill_alpha = 0.8
    # ==============================================

    return p


def update(attr, old, new):
    layout.children[1] = create_figure()


x = Select(title='X-Axis', value='mpg', options=columns)
x.on_change('value', update)

y = Select(title='Y-Axis', value='hp', options=columns)
y.on_change('value', update)

size = Select(title='Size', value='None', options=['None'] + continuous)
size.on_change('value', update)

color = Select(title='Color', value='None', options=['None'] + continuous)
color.on_change('value', update)

controls = column(x, y, color, size, width=200)
layout = row(controls, create_figure())

curdoc().add_root(layout)
curdoc().title = "Crossfilter"

效果说明 #

• 切换X/Y轴字段时，回归线会跟随当前选中的字段自动重新拟合计算，和散点同步更新
• 选中离散分类字段作为X或Y轴时，回归线自动隐藏，不会触发类型错误
• 回归线默认用红色虚线绘制，和原有蓝色散点有明显视觉区分，可自行修改p.line()的颜色、线宽、线型参数调整样式

内容的提问来源于stack exchange，提问作者Javi Nitro 5