seaborn kdeplot报错TypeError：变量类型不符的原因及解决方法

原始代码 #

import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
f = open('/Users/justkiddings/Desktop/Python/TKO/TKO.csv', ) # open the file

data = pd.read_csv(f) # read the .csv file
co=data['CO']
no2=data['NO2']  # set the data

sns.kdeplot(co,shade=True) # to make a figure of CO
sns.kdeplot(no2,shade=True) # to make a figure of NO2
plt.show()

报错信息 #

运行上述代码触发TypeError，错误回溯如下：

runfile('/Users/justkiddings/Desktop/Python/Code/untitled18.py', wdir='/Users/justkiddings/Desktop/Python/Code')
Traceback (most recent call last):

  File "/Users/justkiddings/opt/anaconda3/lib/python3.9/site-packages/spyder_kernels/py3compat.py", line 356, in compat_exec
    exec(code, globals, locals)

  File "/Users/justkiddings/Desktop/Python/Code/untitled18.py", line 11, in <module>
    sns.kdeplot(co,shade=True) # to make a figure of CO

  File "/Users/justkiddings/opt/anaconda3/lib/python3.9/site-packages/seaborn/_decorators.py", line 46, in inner_f
    return f(**kwargs)

  File "/Users/justkiddings/opt/anaconda3/lib/python3.9/site-packages/seaborn/distributions.py", line 1762, in kdeplot
    p._attach(ax, allowed_types=["numeric", "datetime"], log_scale=log_scale)

  File "/Users/justkiddings/opt/anaconda3/lib/python3.9/site-packages/seaborn/_core.py", line 1125, in _attach
    raise TypeError(err)

TypeError: The x variable is categorical, but one of ['numeric', 'datetime'] is required

数据样例 #

DATE,HOUR,STATION,CO,FSP,NO2,NOX,O3,RSP,SO2
1/1/2022,1,TSEUNG KWAN O,67,29,17,18,125,47,4
1/1/2022,2,TSEUNG KWAN O,68,31,18,18,119,50,5
1/1/2022,3,TSEUNG KWAN O,67,35,16,16,116,47,5
1/1/2022,4,TSEUNG KWAN O,66,30,14,15,112,46,4
1/1/2022,5,TSEUNG KWAN O,67,25,14,14,102,47,4

报错原因 #

• sns.kdeplot()是用于绘制连续数值型数据的核密度图，要求输入的x变量必须是数值型（numeric）或日期时间型（datetime）。
• 报错显示CO和NO2列被识别为分类数据类型（categorical），原因通常是：
  1. CSV文件中对应列存在非数值内容（如空值、字符串格式的数字、异常符号等）；
  2. pandas读取数据时自动推断类型出错，将本应是数值的列判定为分类类型。

解决方法 #

1. 强制转换数据类型并处理异常值 #

读取数据后，将CO和NO2列强制转为数值型，同时将非数值内容转为NaN并可选删除：

import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns

# 直接用pd.read_csv读取，无需手动open文件
data = pd.read_csv('/Users/justkiddings/Desktop/Python/TKO/TKO.csv')

# 转换为数值型，errors='coerce'将非数值转为NaN
data['CO'] = pd.to_numeric(data['CO'], errors='coerce')
data['NO2'] = pd.to_numeric(data['NO2'], errors='coerce')

# 可选：删除包含NaN的行（如果存在）
data = data.dropna(subset=['CO', 'NO2'])

co = data['CO']
no2 = data['NO2']

sns.kdeplot(co, shade=True, label='CO')
sns.kdeplot(no2, shade=True, label='NO2')
plt.legend()
plt.show()

2. 排查原始数据 #

打开CSV文件检查CO和NO2列：

• 确认所有行的对应字段都是纯数字，无空单元格、空格、字母（如"NA"）或其他非数字字符；
• 若存在异常值，直接在CSV中修正后再读取。

3. 读取时指定数据类型 #

如果确定数据都是数值，读取时直接指定列类型，避免pandas自动推断错误：

data = pd.read_csv(
    '/Users/justkiddings/Desktop/Python/TKO/TKO.csv',
    dtype={'CO': float, 'NO2': float}
)

内容的提问来源于stack exchange，提问作者Jesse Ko