Python中的groupby是否会产生列或数据框？

Question

Lets say I have a dataframe as

age Late
1     1
2     5
3     48
4     46
5     6

...

I want to replace all values in Late that are 46 or 48 with the median of Late . I believe the command is

trainDF.groupby('Late').transform(getmedian)

however, is the result the whole dataframe trainDF ? or is it just the Late column?

That is, does the below make sense?

trainDF=trainDF.groupby('Late').transform(getmedian)

Or does the below make sense?

newLate = trainDF.groupby('Late').transform(getmedian)

I tried

newLate = trainDF.groupby('Late').transform(getmedian)
newLate.max()

prints out Unnamed: 0

and trainDF['newLate'].max()

prints out KeyError: 'newLate'

If I try

trainDF=trainDF.groupby('Late').transform(getmedian)

and print out

trainDF['Late'].max()

it says KeyError: 'Late'

What am I supposed to do to store the new data frame that replaces the 46 and 48 values in Late ? I ask because I eventually want to make sure this works by printing out the max value of the modified Late column, and make sure it shows 6 (or any number less than 46), not 48

Answer 1

You'e asking a lot of questions here, but I'll address what seems to be the main one:

I want to replace all values in Late that are 46 or 48 with the median of Late.

You can do that this way:

>>> df = pd.DataFrame({'age': [1,2,3,4,5],
                       'Late': [1,5,48,46,6]})
>>> df.loc[df['Late'].isin([46, 48]), 'Late'] = df['Late'].median()
>>> df
   Late  age
0     1    1
1     5    2
2     6    3
3     6    4
4     6    5

Answer 2

Here is a good one liner for you:

trainDF["Late"].loc[(trainDF["Late"] == 48) | (trainDF["Late"] == 46)] = trainDF["Late"].median()

Bear in mind that groupby didn't really apply in your case and that it returns a DataFrame -ish object