Plotly堆叠条形图触发UFuncTypeError，求问题排查与解决方案

我希望基于admin.disease_code中的亚型，使用Plotly创建展示各特征频率的堆叠条形图，但运行代码时触发了UFuncTypeError: ufunc 'add' did not contain a loop with signature matching types (dtype('<U18'), dtype('int64')) -> None错误。

相关代码与报错信息 #

导入代码 #

import pandas as pd
import numpy as np
import plotly.express as px

分组聚合代码 #

agg_df = df.groupby(["admin.disease_code"]).agg({'count'})
agg_df = agg_df.drop("tcga_id", axis=1)
agg_df.columns = agg_df.columns.get_level_values(0)
agg_df.rename(columns = {'patient.days_to_death':'days_to_death', 'patient.vital_status':'vital_status', 'patient.age_at_initial_pathologic_diagnosis':'age_at_initial_pathologic_diagnosis', 'patient.gender':'gender','patient.karnofsky_performance_score':'karnofsky_performance_score'}, inplace = True)

绘图代码 #

fig = px.bar(agg_df, x=agg_df.columns, title="Clinical information for RCC subtypes")
fig.show()

报错回溯 #

---------------------------------------------------------------------------
UFuncTypeError                            Traceback (most recent call last)
/tmp/ipykernel_17/79231555.py in <module>
----> 1 fig = px.bar(agg_df, x=agg_df.columns, title="Clinical information for RCC subtypes")
      2 fig.show()

/opt/conda/lib/python3.7/site-packages/plotly/express/_chart_types.py in bar(data_frame, x, y, color, pattern_shape, facet_row, facet_col, facet_col_wrap, facet_row_spacing, facet_col_spacing, hover_name, hover_data, custom_data, text, base, error_x, error_x_minus, error_y, error_y_minus, animation_frame, animation_group, category_orders, labels, color_discrete_sequence, color_discrete_map, color_continuous_scale, pattern_shape_sequence, pattern_shape_map, range_color, color_continuous_midpoint, opacity, orientation, barmode, log_x, log_y, range_x, range_y, text_auto, title, template, width, height)
    375         constructor=go.Bar,
    376         trace_patch=dict(textposition="auto"),
--> 377         layout_patch=dict(barmode=barmode),
    378     )
    379 

/opt/conda/lib/python3.7/site-packages/plotly/express/_core.py in make_figure(args, constructor, trace_patch, layout_patch)
   1988     apply_default_cascade(args)
   1989 
-> 1990     args = build_dataframe(args, constructor)
   1991     if constructor in [go.Treemap, go.Sunburst, go.Icicle] and args["path"] is not None:
   1992         args = process_dataframe_hierarchy(args)

/opt/conda/lib/python3.7/site-packages/plotly/express/_core.py in build_dataframe(args, constructor)
   1457             value_vars=wide_value_vars,
   1458             var_name=var_name,
-> 1459             value_name=value_name,
   1460         )
   1461         assert len(df_output.columns) == len(set(df_output.columns)), (

/opt/conda/lib/python3.7/site-packages/pandas/core/frame.py in melt(self, id_vars, value_vars, var_name, value_name, col_level, ignore_index)
   8344             value_name=value_name,
   8345             col_level=col_level,
-> 8346             ignore_index=ignore_index,
   8347         )
   8348 

/opt/conda/lib/python3.7/site-packages/pandas/core/reshape/melt.py in melt(frame, id_vars, value_vars, var_name, value_name, col_level, ignore_index)
    142         mdata[col] = id_data
    143 
-> 144     mcolumns = id_vars + var_name + [value_name]
    145 
    146     # error: Incompatible types in assignment (expression has type "ndarray",

UFuncTypeError: ufunc 'add' did not contain a loop with signature matching types (dtype('<U18'), dtype('int64')) -> None

数据信息 #

数据样例（df.head().to_dict()） #

{'admin.disease_code': {'TCGA-2K-A9WE-01A-11D-A381-01': 'kirp',
  'TCGA-2Z-A9J1-01A-11D-A381-01': 'kirp',
  'TCGA-2Z-A9J2-01A-11D-A381-01': 'kirp',
  'TCGA-2Z-A9J3-01A-12D-A381-01': 'kirp',
  'TCGA-2Z-A9J5-01A-21D-A381-01': 'kirp'},
 'patient.days_to_death': {'TCGA-2K-A9WE-01A-11D-A381-01': nan,
  'TCGA-2Z-A9J1-01A-11D-A381-01': nan,
  'TCGA-2Z-A9J2-01A-11D-A381-01': nan,
  'TCGA-2Z-A9J3-01A-12D-A381-01': '1771',
  'TCGA-2Z-A9J5-01A-21D-A381-01': nan},
 'patient.age_at_initial_pathologic_diagnosis': {'TCGA-2K-A9WE-01A-11D-A381-01': '53',
  'TCGA-2Z-A9J1-01A-11D-A381-01': '71',
  'TCGA-2Z-A9J2-01A-11D-A381-01': '71',
  'TCGA-2Z-A9J3-01A-12D-A381-01': '67',
  'TCGA-2Z-A9J5-01A-21D-A381-01': '80'},
 'patient.karnofsky_performance_score': {'TCGA-2K-A9WE-01A-11D-A381-01': nan,
  'TCGA-2Z-A9J1-01A-11D-A381-01': nan,
  'TCGA-2Z-A9J2-01A-11D-A381-01': nan,
  'TCGA-2Z-A9J3-01A-12D-A381-01': nan,
  'TCGA-2Z-A9J5-01A-21D-A381-01': nan},
 'tcga_id': {'TCGA-2K-A9WE-01A-11D-A381-01': 'TCGA-2K-A9WE-01A',
  'TCGA-2Z-A9J1-01A-11D-A381-01': 'TCGA-2Z-A9J1-01A',
  'TCGA-2Z-A9J2-01A-11D-A381-01': 'TCGA-2Z-A9J2-01A',
  'TCGA-2Z-A9J3-01A-12D-A381-01': 'TCGA-2Z-A9J3-01A',
  'TCGA-2Z-A9J5-01A-21D-A381-01': 'TCGA-2Z-A9J5-01A'},
 'survival': {'TCGA-2K-A9WE-01A-11D-A381-01': 'lts',
  'TCGA-2Z-A9J1-01A-11D-A381-01': 'lts',
  'TCGA-2Z-A9J2-01A-11D-A381-01': 'lts',
  'TCGA-2Z-A9J3-01A-12D-A381-01': 'non-lts',
  'TCGA-2Z-A9J5-01A-21D-A381-01': 'lts'}}

agg_df数据类型 #

0
patient.days_to_death                          int64
patient.age_at_initial_pathologic_diagnosis    int64
patient.karnofsky_performance_score            int64
survival                                       int64
dtype: object

期望输出 #

错误原因分析 #

1. 参数与数据结构不匹配：agg_df的索引是疾病亚型，列是各特征的非空计数，但px.bar错误地将列名设为x轴，而非疾病亚型，导致数据格式转换时出现类型冲突。
2. 聚合逻辑偏差：agg({'count'})统计的是各列非空值数量，并非特征的类别频率（如survival的lts/non-lts在各亚型中的占比），不符合堆叠频率图的需求。
3. 数据类型混乱：原始数据中数值型列（如patient.days_to_death）以字符串形式存储，后续聚合后类型冲突，触发运算错误。

分步解决代码 #

1. 数据预处理：修正数据类型 #

将字符串格式的数值列转为数值类型，连续值（如年龄）分箱以便统计频率：

# 转换数值型列
df['patient.days_to_death'] = pd.to_numeric(df['patient.days_to_death'], errors='coerce')
df['patient.age_at_initial_pathologic_diagnosis'] = pd.to_numeric(df['patient.age_at_initial_pathologic_diagnosis'], errors='coerce')
df['patient.karnofsky_performance_score'] = pd.to_numeric(df['patient.karnofsky_performance_score'], errors='coerce')

# 年龄分箱，方便统计频率
df['age_group'] = pd.cut(df['patient.age_at_initial_pathologic_diagnosis'], bins=[0,50,70,100], labels=['<50','50-70','>70'])

2. 统计特征类别频率 #

针对分类特征，统计各疾病亚型下的类别数量：

# 选择需要统计的分类特征
features = ['survival', 'age_group', 'patient.vital_status', 'patient.gender']

# 遍历特征生成频率数据
freq_dfs = []
for feature in features:
    freq_df = df.groupby(['admin.disease_code', feature]).size().reset_index(name='count')
    freq_df['feature'] = feature  # 标记所属特征，用于后续子图展示
    freq_dfs.append(freq_df)

# 合并所有特征的频率数据
final_df = pd.concat(freq_dfs, ignore_index=True)
final_df.rename(columns={'admin.disease_code':'subtype'}, inplace=True)

3. 绘制堆叠条形图 #

使用Plotly生成按亚型分组、特征类别堆叠的条形图：

fig = px.bar(final_df, 
             x='subtype', 
             y='count', 
             color=final_df.columns[1],  # 自动匹配特征类别列
             facet_col='feature', 
             barmode='stack',
             title="RCC亚型临床信息特征频率堆叠图",
             labels={'count':'样本数量', 'subtype':'疾病亚型'})

# 调整布局，优化展示效果
fig.update_layout(
    facet_col_wrap=2,  # 每行显示2个子图
    height=600,
    width=1000
)

fig.show()

完整可运行代码 #

import pandas as pd
import numpy as np
import plotly.express as px

# 构造示例数据（替换为你的原始数据即可）
df = pd.DataFrame({
    'admin.disease_code': ['kirp','kirp','kirp','kirp','kirp','kich','kich','kich'],
    'patient.days_to_death': [np.nan, np.nan, np.nan, '1771', np.nan, '980', np.nan, '1200'],
    'patient.age_at_initial_pathologic_diagnosis': ['53','71','71','67','80','45','62','75'],
    'patient.karnofsky_performance_score': [np.nan]*8,
    'tcga_id': ['TCGA-2K-A9WE-01A','TCGA-2Z-A9J1-01A','TCGA-2Z-A9J2-01A','TCGA-2Z-A9J3-01A','TCGA-2Z-A9J5-01A','TCGA-3K-A9WX-01A','TCGA-3Z-A9J6-01A','TCGA-4Z-A9J7-01A'],
    'survival': ['lts','lts','lts','non-lts','lts','non-lts','lts','non-lts'],
    'patient.vital_status': [np.nan, 'Alive', np.nan, 'Dead', np.nan, 'Dead', 'Alive', 'Dead'],
    'patient.gender': ['Male','Female','Male','Male','Female','Female','Male','Female']
})

# 数据预处理
df['patient.days_to_death'] = pd.to_numeric(df['patient.days_to_death'], errors='coerce')
df['patient.age_at_initial_pathologic_diagnosis'] = pd.to_numeric(df['patient.age_at_initial_pathologic_diagnosis'], errors='coerce')
df['patient.karnofsky_performance_score'] = pd.to_numeric(df['patient.karnofsky_performance_score'], errors='coerce')
df['age_group'] = pd.cut(df['patient.age_at_initial_pathologic_diagnosis'], bins=[0,50,70,100], labels=['<50','50-70','>70'])

# 统计特征频率
features = ['survival', 'age_group', 'patient.vital_status', 'patient.gender']
freq_dfs = []
for feature in features:
    freq_df = df.groupby(['admin.disease_code', feature]).size().reset_index(name='count')
    freq_df['feature'] = feature
    freq_dfs.append(freq_df)

final_df = pd.concat(freq_dfs, ignore_index=True)
final_df.rename(columns={'admin.disease_code':'subtype'}, inplace=True)

# 绘制堆叠条形图
fig = px.bar(final_df, 
             x='subtype', 
             y='count', 
             color=final_df.columns[1],
             facet_col='feature', 
             barmode='stack',
             title="RCC亚型临床信息特征频率堆叠图",
             labels={'count':'样本数量', 'subtype':'疾病亚型'})

fig.update_layout(
    facet_col_wrap=2,
    height=600,
    width=1000
)

fig.show()

内容的提问来源于stack exchange，提问作者melolilili