python - 字典列表列表到 pandas DataFrame

Question

我正在尝试拟合这些数据:

[['Manufacturer: Hyundai',
  'Model: Tucson',
  'Mileage: 258000 km',
  'Registered: 07/2019'],
 ['Manufacturer: Mazda',
  'Model: 6',
  'Year: 2014',
  'Registered: 07/2019']]

到 pandas DataFrame。

并非所有标签都出现在每条记录中，例如，一些记录有“里程”，而另一些则没有，反之亦然。我总共有 26 个特征，很少有项目拥有所有这些特征。

我想构建将在列中保存特征的 pandas DataFrame，如果特征不存在，则内容应为“NaN”。

我有

colnames=['Manufacturer', 'Model', 'Mileage', 'Registered', 'Year'...(all 26 features here)] 
df = pd.read_csv("./data/output.csv", sep=",", names=colnames, header=None)

很少有第一个先决条件列会给出预期的输出，但是当涉及到可选功能时，而不是缺少数据导致之后的功能出现在错误的列下。仅当所有功能都存在时，记录才能正确映射。

我忘了提到一些缺失值的特征也没有“:”但存在于列表中。所以在这两种情况下:

• “里程”，(缺少值，但也缺少“:”)
• 总记录中缺少“里程”

两种情况的赋值都应为“NaN”。

Answer 1

对字典列表使用嵌套列表理解并传递给 DataFrame 构造函数，如果缺少相同的键则添加 NaN:

L = [['Manufacturer: Hyundai',
  'Model: Tucson',
  'Mileage: 258000 km',
  'Registered: 07/2019'],
 ['Manufacturer: Mazda',
  'Model: 6',
  'Year: 2014',
  'Registered: 07/2019']]

df = pd.DataFrame([dict(y.split(':') for y in x) for x in L])
print (df)
  Manufacturer     Mileage    Model Registered   Year
0      Hyundai   258000 km   Tucson    07/2019    NaN
1        Mazda         NaN        6    07/2019   2014

编辑:您可以使用 .split(maxsplit=1) 按第一个空格拆分:

L = [['Manufacturer Hyundai',
  'Model Tucson',
  'Mileage 258000 km',
  'Registered 07/2019'],
 ['Manufacturer Mazda',
  'Model 6',
  'Year 2014',
  'Registered 07/2019']]


df = pd.DataFrame([dict(y.split(maxsplit=1) for y in x) for x in L])
print (df)

  Manufacturer    Mileage   Model Registered  Year
0      Hyundai  258000 km  Tucson    07/2019   NaN
1        Mazda        NaN       6    07/2019  2014

编辑:

L = [['Manufacturer  Hyundai',
  'Model  Tucson',
  'Mileage  258000 km',
  'Registered  07/2019'],
 ['Manufacturer  Mazda',
  'Model  6',
  'Year  2014',
  'Registered  07/2019',
  'Additional equipment aaa']]

words2 = ['Additional equipment']

L1 = []
for x in L:
    di = {}
    for y in x:
        for word in words2:
            if set(word.split(maxsplit=2)[:2]) < set(y.split()):
                i, j, k = y.split(maxsplit=2)
                di['_'.join([i, j])] = k
            else:
                i, j = y.split(maxsplit=1)
                di[i] = j
    L1.append(di)

df = pd.DataFrame(L1)
print (df)
  Additional_equipment Manufacturer    Mileage   Model Registered  Year
0                  NaN      Hyundai  258000 km  Tucson    07/2019   NaN
1                  aaa        Mazda        NaN       6    07/2019  2014