【发布时间】：2020-07-28 12:07:31
【问题描述】：
我有a dataframe，我想将部分行转换为多行。实际上，这些行代表Questions 列中的一个问题，而Answer_i 列中的这些问题的答案。例如以下行：
QID Questions QType Answer_1 Answer_2 Answer_3 Answer_4 Answer_5 Answer_6 Answer_7 Answer_8 Answer_9 Answer_10 Answer_11 Answer_12 Answer_13 Answer_14 Answer_15
1177 The travel restrictions of COVID-19 have been ... Likert Scale Very important consideration Important consideration Somewhat consider Not an important consideration Do not consider Discounted flights Very important consideration Important consideration Somewhat consider Not an important consideration Do not consider Baggage policy Very important consideration Important consideration Somewhat consider Not an important considera... Do not consider
我想为这一行获取以下数据框：
QID Questions QType Answer_1 Answer_2 Answer_3 Answer_4 ...
1263 1177 The travel restrictions of COVID-19 have been lifted and you are looking to book a flight. To what extent are the following factors considerations in your choice of flight? Likert Scale
Very important consideration Important consideration Somewhat consider Not an important consideration Do not consider
1264 1177_1 Discounted flights Likert Scale Very important consideration Important consideration Somewhat consider Not an important consideration Do not consider
1265 1177_2 Baggage policy Likert Scale Very important consideration Important consideration Somewhat consider Not an important consideration Do not consider
到目前为止，我一直在尝试迭代答案：
for i, row in df.iterrows():
passed_items = []
for cell in row:
if cell in passed_items:
print("need to create a new line")
answers = {f"Answer{i}": passed_items[i] for i in range(0, len(passed_items))} # dyanmically allocate to place them in the right columns
dict_replacing = {'Questions': questions, **answers} # dictionary that will replace the forle create the new lines
df1 = pd.DataFrame(dict_replacing)
df = df1.combine_first(df)
passed_items = []
passed_items.append(str(cell))
但它给了我回报：
Answer_2 Answer_9 Answer_0 Answer_1 Answer_10 Answer_11 Answer_12 Answer_13 Answer_14 Answer_2 Answer_3 Answer_4 Answer_5 Answer_6 Answer_7 Answer_8 QID QType Questions
0 NaN NaN Very important consideration Important consideration NaN NaN NaN NaN NaN Somewhat consider Not an important consideration Do not consider Baggage policy Discounted flights NaN NaN NaN NaN The airline/company you fly with
1 NaN NaN Very important consideration Important consideration NaN NaN NaN NaN NaN Somewhat consider Not an important consideration Do not consider Baggage policy Discounted flights NaN NaN NaN NaN The departure airport
2 NaN NaN Very important consideration Important consideration NaN NaN NaN NaN NaN Somewhat consider Not an important consideration Do not consider Baggage policy Discounted flights NaN NaN NaN NaN Duration of flight/route
3 NaN NaN Very important consideration Important consideration NaN NaN NaN NaN NaN Somewhat consider Not an important consideration Do not consider Baggage policy Discounted flights NaN NaN NaN NaN Price
4 NaN NaN Very important consideration Important consideration NaN NaN NaN NaN NaN Somewhat consider Not an important consideration Do not consider Baggage policy Discounted flights NaN NaN NaN NaN Baggage policy
5 NaN NaN Very important consideration Important consideration NaN NaN NaN NaN NaN Somewhat consider Not an important consideration Do not consider Baggage policy Discounted flights NaN NaN NaN NaN Environmental impacts
1263 Important consideration Somewhat consider No... Do not consider NaN Very important consideration Baggage policy Very important consideration Important consideration Somewhat consider Not an important considera... Do not consider NaN Do not consider Discounted flights Very important consideration Important consideration Somewhat consider Not an important consideration 1177.0 Likert Scale The travel restrictions of COVID-19 have been ...
不遵守列的顺序，其中一些是双重的。
更新
我一直在努力理解 Rob Raymond 的回答。
我不明白：
- for 循环：
for i in range(3, len(r)-len(repeat)):我们是否对每一列进行迭代，直到数据帧的最后一个？ - 爆炸函数
df.apply(lambda r: getquestions(r), axis=1).explode("Questions"):
根据w3ressource：
explode() 函数用于将列表中的每个元素转换为一行，复制索引值。
是这个东西把 r 变成了我要找的列表吗？
- 它如何创建一个新的数据报。我知道这与前面的答案有关。
这是代码，我的 cmets：
import collections
df = pd.DataFrame({"QID":[1177],"Questions":["The travel restrictions of COVID-19 have been lifted and you are looking to book a flight. To what extent are the following factors considerations in your choice of flight?"],"QType":["Likert Scale"],"Answer0":["Very important consideration"],"Answer1":["Important consideration"],"Answer2":["Somewhat consider"],"Answer3":["Not an important consideration"],"Answer4":["Do not consider"],"Answer5":["Discounted flights"],"Answer6":["Very important consideration"],"Answer7":["Important consideration"],"Answer8":["Somewhat consider"],"Answer9":["Not an important consideration"],"Answer10":["Do not consider"],"Answer11":["Baggage policy"],"Answer12":["Very important consideration"],"Answer13":["Important consideration"],"Answer14":["Somewhat consider"],"Answer15":["Not an important consideration"],"Answer16":["Do not consider"],"Answer17":["Price of flights"],"Answer18":["Very important consideration"],"Answer19":["Important consideration"],"Answer20":["Somewhat consider"],"Answer21":["Not an important consideration"],"Answer22":["Do not consider"],"Answer23":["Insurance"],"Answer24":["Very important consideration"],"Answer25":["Important consideration"],"Answer26":["Somewhat consider"],"Answer27":["Not an important consideration"],"Answer28":["Do not consider"],"Answer29":["Airport services"],"Answer30":["Very important consideration"],"Answer31":["Important consideration"],"Answer32":["Somewhat consider"],"Answer33":["Not an important consideration"],"Answer34":["Do not consider"],"Answer35":["Environmental impact"],"Answer36":["Very important consideration"],"Answer37":["Important consideration"],"Answer38":["Somewhat consider"],"Answer39":["Not an important consideration"],"Answer40":["Do not consider"],"Answer41":["In-flight service"],"Answer42":["Very important consideration"],"Answer43":["Important consideration"],"Answer44":["Somewhat consider"],"Answer45":["Not an important consideration"],"Answer46":["Do not consider"],"Answer47":["Customer support"],"Answer48":["Very important consideration"],"Answer49":["Important consideration"],"Answer50":["Somewhat consider"],"Answer51":["Not an important consideration"],"Answer52":["Do not consider"],"Answer53":["Overcrowding on aircraft/airports"],"Answer54":["Very important consideration"],"Answer55":["Important consideration"],"Answer56":["Somewhat consider"],"Answer57":["Not an important consideration"],"Answer58":["Do not consider"],"Answer59":["Airport safety after COVID-19"],"Answer60":["Very important consideration"],"Answer61":["Important consideration"],"Answer62":["Somewhat consider"],"Answer63":["Not an important consideration"],"Answer64":["Do not consider"],"Answer65":["Refund policy"]})
def getquestions(r):
# counter
repeat = list({k:v for k,v in collections.Counter(r[3:].values).items() if v>1}) # get all the questions
questions = []
firstfound = 0
#
for i in range(3, len(r)-len(repeat)): # I don't get this one
if r[i:i+len(repeat)].tolist()==repeat: # I think we are trying to get the subset that are repeat
if r[i+len(repeat):i+len(repeat)+1].values[0] is not None: # here we get the question
questions.append(r[i+len(repeat):i+len(repeat)+1].values[0]) # we store it
if firstfound==0: firstfound = i+len(repeat) # so when it's not 0, we do not update? Why? Why is this thing for?
if len(questions) > 0: #weird cases ?
# somethong odd, sometimes it's a list other times a str
newq = r[1] + questions if isinstance(r[1], list) else [r[1]] + questions
r[1] = newq
# reset all the questions that have been used by list
for i in range(firstfound, len(r)):
if isinstance(r[i], str): r[i] = None
return r
def fixqid(c):
return [id if i==0 or c[i-1]!=id else f"{id}_{i}" for i, id in enumerate(c)]
df = df.apply(lambda r: getquestions(r), axis=1).explode("Questions").reset_index().drop("index", 1) # what does explode stands for?
df["QID"] = fixqid(df["QID"].values)
df
【问题讨论】：
-
df = pd.read_csv('DataFrame.csv')。这对我来说很好
-
@RahulVishwakarma 它可能会读取数据帧，但可能不会像我上面显示的那样将某些行的一部分转换为几行
-
它对我来说是正确导入的，也许你的电脑有问题，而不是你的代码
-
你是如何读取数据框的？显示代码
-
@RahulVishwakarma 我没有，我正在尝试上面给出的摘录，索引 1263
标签： python python-3.x pandas dataframe replace