使用来自另一个数据帧的条件组填充pandas数据帧

Question

I have two dataframes, one being a list of teams and scores sorted by date, and the second individual players with the date. I have 60 columns of matching stats in these dataframes, and I am trying to have a code that replaces the values in each column in df2 with the average the opponent had in the dates previous from df1:

df1:                         df2:
   date        team  scr        name           team opp  date        scr
0  2016-04-03  KCR   5.70    0  Erasmo Ramirez TBR  TOR  2016-04-06  7.90
1  2016-04-03  NYM   4.70    1  Erasmo Ramirez TBR  BAL  2016-04-10  1.30
2  2016-04-03  PIT   6.30    2  Erasmo Ramirez TBR  CLE  2016-04-13  9.30
3  2016-04-03  STL   3.40    etc...
4  2016-04-03  TBR   4.80
5  2016-04-03  TOR   6.20*
6  2016-04-04  ARI   7.40
7  2016-04-04  ATL   5.30
8  2016-04-04  BAL   7.00
9  2016-04-04  CHC   9.60
10 2016-04-04  TOR   7.50*
etc...

So in this example the first entry under 'scr' in df2 would be changed from 7.90 to 6.85 as that is the average scr for TOR for the dates leading up to 4-6 (4-3 and 4-4)

I tried the following (and other similar options) and had no luck:

jf = df1.groupby('team')
df2['scr'] = jf.apply(lambda x: x[(df1['date']<x['date'])&(df1['team']==x['opp'])]['scr'].sum())

ValueError: Series lengths must match to compare

Any solutions? And also is there a possible way to iterate over all the columns with just one block of code, or do I have to have code for each column?

Answer 1

I found one solution, perhaps not very elegant, and I'll check back in case anyone actually sees this and has a better one. I created a list of the column headers, then used iterrows to iterate through both each column and row in the column:

    batcol = pd.DataFrame(df1.columns)
    batcol = batcol.iloc[6:-1]
    batcol = batcol.reset_index(drop=True)
    for index, row in batcol.iterrows():
        for i, rows in df2.iterrows():
            df2.loc[i, row] = df1[((df1['date'] < rows['date']) & (df1['team'] == rows['opp'])) == True].mean()[row].sum()