如何根据Classification列值为Plotly垂直矩形着色？

尝试用Plotly创建包含散点图和垂直矩形（表示事件时长）的图表，希望根据Classification列给矩形设置不同颜色，但循环中设置fillcolor时出现KeyError: 'Classification'；固定颜色为'red'时可正常显示，但所有矩形颜色一致。

数据结构 #

df1

Datetime, Frequency_hz
15/08/2023 09:09:22, 1575422000
15/08/2023 10:10:32, 1575452000
15/08/2023 10:10:35, 1572322000
15/08/2023 11:11:54, 1575432000

events_df

Date_Start_Time, Date_End_Time, Classification, Band
15/08/2023 09:23:22,15/08/2023 09:35:12, Class 1,A
15/08/2023 09:43:40,15/08/2023 09:50:47, Class 2,A
15/08/2023 10:13:44,15/08/2023 10:15:32, Class 3,A
15/08/2023 11:33:11,15/08/2023 11:39:23, Class 4,B

报错代码 #

# Loading data
DATA = 'C:\work\20230815_UTC.csv'
EVENT_DATA = r"C:\work\events.csv"
BAND = 'A'

df = pd.read_csv(DATA)
events_df = pd.read_csv(EVENT_DATA)

# Filter for band
events_df = events_df[(events_df['Band'] == BAND)]

# Filter data for specific weekday
events_df = events_df.loc[events_df['Date_Start_time'].dt.weekday == 1] # 1= Tues, 2=Wed, 3=Thurs


df = pd.read_csv(DATA)
# Create plots 
fig1 = go.Figure()
fig1.add_scatter(
        x=df1.Datetime,
        y=df1.Frequency_hz,
        mode="markers",
        name="Points",
        showlegend=True
    )

# Dictionary for colouring classes
colours = {'Class 1':'red','Class 2':'violet',
                'Class 3':'aqua','Class 4':'brown'}

# Loop through events table and plot

for i in events_df.index:
    start_time = events_df.loc[i,"Date_Start_time"]
    end_time = events_df.loc[i,"Date_End_time"]
    colour = events_df.loc[i,"Classification"]
    fig1.add_vrect(x0=start_time,
                  x1=end_time,
                  fillcolor=colours['Classification'],
                  opacity=0.5,
                  name='Events'
                  )

# Update various layout parameters
fig1.update_layout(
    title=f"Events on 15th August 2023",
    xaxis_title="Time (UTC)",
    yaxis_title="Frequency (Hz)",
    font=dict(
        family="Courier New, monospace",
        size=18,
        color="RebeccaPurple"
    )
)
fig1.show()

1. 修复KeyError核心问题 #

报错原因是你在fillcolor=colours['Classification']中用字符串'Classification'作为字典键，但颜色字典colours的键是'Class 1'这类分类值，而非列名。你已经获取了当前行的分类值并存入colour变量，直接用该变量作为键即可：

fillcolor=colours[colour],

或直接读取当前行的分类值：

fillcolor=colours[events_df.loc[i,"Classification"]],

2. 修复日期类型转换问题 #

代码中直接调用events_df['Date_Start_time'].dt.weekday会报错，因为pd.read_csv默认将日期列读取为字符串，需先转换为datetime类型；同时注意原数据列名是Date_Start_Time（大写T），代码中误用了Date_Start_time（小写t），需统一：

# 转换日期列为datetime类型，统一列名
events_df['Date_Start_time'] = pd.to_datetime(events_df['Date_Start_Time'])
events_df['Date_End_time'] = pd.to_datetime(events_df['Date_End_Time'])
df['Datetime'] = pd.to_datetime(df['Datetime'])

3. 其他优化点 #

• 移除重复的df = pd.read_csv(DATA)代码；
• 修正散点图中变量名不一致问题（代码中定义df但使用df1）；
• 将矩形的name设置为当前分类值，让图例区分不同类别的事件。

修改后的完整代码 #

import pandas as pd
import plotly.graph_objects as go

# Loading data
DATA = r'C:\work\20230815_UTC.csv'  # 加r避免路径转义问题
EVENT_DATA = r"C:\work\events.csv"
BAND = 'A'

df = pd.read_csv(DATA)
events_df = pd.read_csv(EVENT_DATA)

# 转换日期列为datetime类型，统一列名
events_df['Date_Start_time'] = pd.to_datetime(events_df['Date_Start_Time'])
events_df['Date_End_time'] = pd.to_datetime(events_df['Date_End_Time'])
df['Datetime'] = pd.to_datetime(df['Datetime'])

# Filter for band
events_df = events_df[(events_df['Band'] == BAND)]

# Filter data for specific weekday
events_df = events_df.loc[events_df['Date_Start_time'].dt.weekday == 1] # 1= Tues, 2=Wed, 3=Thurs

# Create plots 
fig1 = go.Figure()
fig1.add_scatter(
        x=df.Datetime,
        y=df.Frequency_hz,
        mode="markers",
        name="Points",
        showlegend=True
    )

# Dictionary for colouring classes
colours = {'Class 1':'red','Class 2':'violet',
                'Class 3':'aqua','Class 4':'brown'}

# Loop through events table and plot
for i in events_df.index:
    start_time = events_df.loc[i,"Date_Start_time"]
    end_time = events_df.loc[i,"Date_End_time"]
    colour = events_df.loc[i,"Classification"]
    fig1.add_vrect(x0=start_time,
                  x1=end_time,
                  fillcolor=colours[colour],
                  opacity=0.5,
                  name=colour  # 图例显示分类名称
                  )

# Update various layout parameters
fig1.update_layout(
    title=f"Events on 15th August 2023",
    xaxis_title="Time (UTC)",
    yaxis_title="Frequency (Hz)",
    font=dict(
        family="Courier New, monospace",
        size=18,
        color="RebeccaPurple"
    )
)
fig1.show()

内容的提问来源于stack exchange，提问作者cappo