pandas-为丢失的记录填写默认值

在制作数据帧之前,我会将缺少值的条目添加到数据中,这样它就已经有了正确的行数。然后,您可以通过将带有默认值的字典传递给 df.fillna() :

import pandas as pd

data = [
 {'shape': 'circle', 'height': 5, 'width': 4},
 {'shape': 'circle', 'height': 2, 'width': 3},
 {'shape': 'square', 'height': 6, 'width': 5}
]

shapes = ['circle', 'square', 'triangle']

shapes_present = {d['shape'] for d in data}
shapes_missing = set(shapes) - shapes_present

for shape in shapes_missing:
    data.append({'shape': shape})
    
df = pd.DataFrame(data)
df.fillna({'height': 10, 'width': 5}, inplace=True)
df

    shape       height  width
0   circle      5.0     4.0
1   circle      2.0     3.0
2   square      6.0     5.0
3   triangle    10.0    5.0

您可以使用默认值制作另一个df,类似于

defaults = pd.DataFrame({'shape': shapes, 'height': 10, 'width': 5})

# result
      shape  height  width
0    circle      10      5
1    square      10      5
2  triangle      10      5

然后将原始df与 defaults 尚未出现在 df :

pd.concat([df, defaults[~defaults['shape'].isin(df['shape'])]], 
          ignore_index=True)

# result
      shape  height  width
0    circle       5      4
1    circle       2      3
2    square       6      5
3  triangle      10      5