问题
I have question relate in pandas dataframe merge.
Plz, below data..
Rating csv
UserID ContentID Rating
U-1 C-1 3
U-1 C-2 4
U-3 C-3 1
U-5 C-1 5
Content csv
Title ContentID Language
T-1 C-1 EN
T-2 C-2 EN
T-3 C-3 EN
User csv
UserID Age Gender
U-1 10 1
U-2 20 0
U-3 30 1
U-4 40 0
U-5 50 1
U-6 60 0
U-7 70 1
I want Result
UserID ContentID Rating Title Language Age Gender
U-1 C-1 3 T-1 EN 10 1
U-1 C-2 4 T-2 EN 10 1
U-1 C-3 NAN T-3 EN 10 1
U-2 C-1 NAN T-1 EN 20 0
U-2 C-2 NAN T-2 EN 20 0
U-2 C-3 NAN T-3 EN 20 0
U-3 C-1 NAN T-1 EN 30 1
U-3 C-2 NAN T-2 EN 30 1
U-3 C-3 1 T-3 EN 30 1
U-4 C-1 NAN T-1 EN 40 0
U-4 C-2 NAN T-2 EN 40 0
U-4 C-3 NAN T-3 EN 40 0
U-5 C-1 5 T-1 EN 50 1
U-5 C-2 NAN T-2 EN 50 1
U-5 C-3 NAN T-3 EN 50 1
U-6 C-1 NAN T-1 EN 60 0
U-6 C-2 NAN T-2 EN 60 0
U-6 C-3 NAN T-3 EN 60 0
U-7 C-1 NAN T-1 EN 70 1
U-7 C-2 NAN T-2 EN 70 1
U-7 C-3 NAN T-3 EN 70 1
Total DF Rows Size are UserID(User csv) Count * ContentID(Content csv) Count ( ex> Above 7 * 3 -> 21 rows)
All DataFrame are relate. - Rating / Content -> ContentID - Rating / User -> UserID
In other words, Result DataFrame is only remain rating zone(NAN), Other zone is none nan.
Real Size Content( 6000 ), User(220000 ) -> Total Result Rows Count : about 1300000000
I try it, but it's raise memoryError...
plz, help me..Thanks..
回答1:
You can use cross join with left join - necessary unique values in df2.ContentID and df3.UserID:
df = pd.merge(pd.merge(df3.assign(A=1), df2.assign(A=1), on='A'), df1, 'left').drop('A', 1)
print (df)
UserID Age Gender Title ContentID Language Rating
0 U-1 10 1 T-1 C-1 EN 3.0
1 U-1 10 1 T-2 C-2 EN 4.0
2 U-1 10 1 T-3 C-3 EN NaN
3 U-2 20 0 T-1 C-1 EN NaN
4 U-2 20 0 T-2 C-2 EN NaN
5 U-2 20 0 T-3 C-3 EN NaN
6 U-3 30 1 T-1 C-1 EN NaN
7 U-3 30 1 T-2 C-2 EN NaN
8 U-3 30 1 T-3 C-3 EN 1.0
9 U-4 40 0 T-1 C-1 EN NaN
10 U-4 40 0 T-2 C-2 EN NaN
11 U-4 40 0 T-3 C-3 EN NaN
12 U-5 50 1 T-1 C-1 EN 5.0
13 U-5 50 1 T-2 C-2 EN NaN
14 U-5 50 1 T-3 C-3 EN NaN
15 U-6 60 0 T-1 C-1 EN NaN
16 U-6 60 0 T-2 C-2 EN NaN
17 U-6 60 0 T-3 C-3 EN NaN
18 U-7 70 1 T-1 C-1 EN NaN
19 U-7 70 1 T-2 C-2 EN NaN
20 U-7 70 1 T-3 C-3 EN NaN
来源：https://stackoverflow.com/questions/46441328/pandas-merge-multi-dataframe-relate-dataframe