使用錯誤列中的某些單元格值修復 pandas DataFrame 的最佳方法

Question

我有一個 pandas DataFrame ，其中一些單元格位於錯誤的列中。 它看起來像這樣：

城市	河	價格	車
紐約	哈德宗	低的	大眾
洛杉磯	育空地區	高的	特斯拉
拉斯維加斯	低的	哈德宗	大眾
低的	紐約	特斯拉	育空地區

我列舉了 City、River、Price 和 Car 列中所有可能值的列表。

修復此數據集並將所有值放在 pandas 的右列中的最佳方法是什么？

更新：

這就是我創建的代碼，對我有用：

x = pd.DataFrame({'name':['andrew', 'karl', 'jhon', 'jack', 'bob', 'high'], 'education': ['high', 'middle', 'lviv', 'high', 'elementary', 'kyiv'], 'city':['lviv', 'kyiv', 'elementary', 'kharkiv', 'kyiv', 'mike']})

ydata = pd.DataFrame(x)

fdict = {}
fdict['name'] = ['andrew', 'karl', 'jhon', 'jack', 'bob', 'mike']
fdict['education'] = ['high', 'middle', 'elementary']
fdict['city'] = ['lviv', 'kyiv', 'kharkiv']

filter = {}
for key in fdict:
    filter[key] = {}

for key1 in fdict:
    regexx = '|'.join(fdict[key1])
    for key2 in fdict:
        filter[key2][key1] = ydata[key2].str.contains(regexx, regex=True, na=False)


indata = {}
for key in fdict:
    indata[key] = {}

for key1 in fdict:
    for key2 in fdict:
        if key1 != key2:
            indata[key1][key2] = ydata.loc[filter[key1][key2], key1]

for key in fdict:
    fn = ~filter[key][key]
    ydata.loc[fn, key] = np.nan


for key1 in fdict:
    for key2 in fdict:
        if key1 != key2:
            ydata.loc[filter[key1][key2], key2] = indata[key1][key2]

Answer 1

這就是我創建的代碼，對我有用：

x = pd.DataFrame({'name':['andrew', 'karl', 'jhon', 'jack', 'bob', 'high'], 'education': ['high', 'middle', 'lviv', 'high', 'elementary', 'kyiv'], 'city':['lviv', 'kyiv', 'elementary', 'kharkiv', 'kyiv', 'mike']})

ydata = pd.DataFrame(x)

fdict = {}
fdict['name'] = ['andrew', 'karl', 'jhon', 'jack', 'bob', 'mike']
fdict['education'] = ['high', 'middle', 'elementary']
fdict['city'] = ['lviv', 'kyiv', 'kharkiv']

filter = {}
for key in fdict:
    filter[key] = {}

for key1 in fdict:
    regexx = '|'.join(fdict[key1])
    for key2 in fdict:
        filter[key2][key1] = ydata[key2].str.contains(regexx, regex=True, na=False)


indata = {}
for key in fdict:
    indata[key] = {}

for key1 in fdict:
    for key2 in fdict:
        if key1 != key2:
            indata[key1][key2] = ydata.loc[filter[key1][key2], key1]

for key in fdict:
    fn = ~filter[key][key]
    ydata.loc[fn, key] = np.nan


for key1 in fdict:
    for key2 in fdict:
        if key1 != key2:
            ydata.loc[filter[key1][key2], key2] = indata[key1][key2]

Answer 2

其他解決方案：

cols_list = []
for col in ydata.columns:
  ydata[col] = ydata[col].apply(lambda x: {key: x for key in fdict.keys() if x in fdict[key]})
ydata_list = ydata.to_dict('split')['data']

# dataset reconstraction
update_ydata_list = []
for y in ydata_list:
  temp_dict = {}
  for d in y:
      temp_dict.update(d)
  update_ydata_list.append(temp_dict.copy())

# final result
pd.DataFrame(update_ydata_list)

您可以在完整數據集上對其進行測試