Python Plotly分组柱状图过滤周末计算日度差值报错排查

• 对pandas DataFrame做布尔条件过滤后，原始行索引会被保留，不会自动重置为从0开始的连续整数。样例数据过滤掉周末行后，剩余行的索引为0和3，但原差值计算逻辑用while循环从0开始按连续整数取y1[i]、y2[i]，当i取到1时，会因为找不到对应索引值触发KeyError。
• 原写法还存在隐性bug：如果两个DataFrame过滤后日期顺序、数量不一致，手动按位置遍历计算的差值会和日期错位，标注结果错误。

只需要在过滤后重置索引，同时用pandas原生向量运算代替手动while循环计算差值，从写法上避免索引遍历错误：

import plotly.graph_objects as go
import pandas as pd

df1 = pd.DataFrame({
   'score': [11,42,31,22],
   'dates': ['7/1/2022','7/2/2022','7/3/2022','7/4/2022'],
   'weekend': [0,1,1,0],
})

df2 = pd.DataFrame({
   'score': [44,45,26,22],
   'dates': ['7/1/2022','7/2/2022','7/3/2022','7/4/2022'],
   'weekend': [0,1,1,0],
})

# 过滤周末数据后重置索引，丢弃原索引
df1 = df1[df1['weekend'] == 0].reset_index(drop=True)
df2 = df2[df2['weekend'] == 0].reset_index(drop=True)

y1 = df1['score']
y2 = df2['score']
# 直接做向量运算计算差值，无需手动写循环
c = (y1 - y2).astype(str).tolist()

fig = go.Figure(data=[
    go.Bar(name='SF Zoo', x=df1['dates'], y=y1),
    go.Bar(name='LA Zoo', x=df2['dates'], y=y2, text=c, textposition='outside')
])

fig.update_layout(barmode='group')
fig.show()

为了避免两个表分别过滤可能出现的日期不匹配、错位问题，可以先按日期合并两个数据集，再统一做过滤、差值计算，适配更复杂的真实数据场景：

import plotly.graph_objects as go
import pandas as pd

df1 = pd.DataFrame({
   'score': [11,42,31,22],
   'dates': ['7/1/2022','7/2/2022','7/3/2022','7/4/2022'],
   'weekend': [0,1,1,0],
})

df2 = pd.DataFrame({
   'score': [44,45,26,22],
   'dates': ['7/1/2022','7/2/2022','7/3/2022','7/4/2022'],
   'weekend': [0,1,1,0],
})

# 按日期关联合并两个数据源的得分
merged_df = pd.merge(
    df1[['dates', 'score', 'weekend']].rename(columns={'score':'sf_score'}),
    df2[['dates', 'score']].rename(columns={'score':'la_score'}),
    on='dates',
    how='inner'
)
# 统一过滤周末数据
merged_df = merged_df[merged_df['weekend'] == 0]
# 计算两组得分差值
merged_df['diff'] = (merged_df['sf_score'] - merged_df['la_score']).astype(str)

fig = go.Figure(data=[
    go.Bar(name='SF Zoo', x=merged_df['dates'], y=merged_df['sf_score']),
    go.Bar(name='LA Zoo', x=merged_df['dates'], y=merged_df['la_score'], text=merged_df['diff'], textposition='outside')
])

fig.update_layout(barmode='group')
fig.show()

内容的提问来源于stack exchange，提问作者tenebris silentio