英文:
Rolling average Pandas for 3 week period for specific column values
问题
我有一个名为'qbPast'的数据框，其中包含了一赛季的NFL球员数据。
我尝试创建一个新的滚动平均，基于"Points"列，对于每个个体球员，每3周一次，前两周应返回该周的得分，之后应返回3周滚动期间的平均值，例如球员A得分20,30,40,30,40，平均值应返回20,30,30,33.3等。
我的尝试如下：# qbPast['Avg'] = qbPast.groupby('Player')['Points'].rolling(3).mean().reset_index(drop=True)
问题是它只返回了所有球员的3周平均值，我需要按球员筛选，以便为每个球员返回滚动平均值，其他球员不应影响滚动平均值。
英文:
I have a dataframe 'qbPast' which contains nfl player data for a season.
P Player Week Team Opp Opp Rank Points Def TD Def INT Def Yds/att Year
2 QB Kyler Murray 2 ARI MIN 14 38.10 1.8125 1.0000 6.9 2021
3 QB Lamar Jackson 2 BAL KC 6 37.26 1.6875 0.9375 7 2021
5 QB Tom Brady 2 TB ATL 28 30.64 1.9375 0.7500 6.8 2021
I am attempting to create a new rolling average based on the "Points" column for each individual player for each 3 week period, for the first two weeks it should just return the points for that week and after that it should return the average for the 3 week moving period e,g Player A scores 20,30,40,30,40 the average should return 20,30,30,33.3 etc.
My attempt # qbPast['Avg'] = qbPast.groupby('Player')['Points'].rolling(3).mean().reset_index(drop=True)
The problem is it is only returning the 3 week average for all players I need it to filter by player so that it returns the rolling average for each player, the other players should not affect the rolling average.
答案1
得分: 1
问题在于索引对齐。不要使用 reset_index，而是删除第零索引级，然后将值分配给一列。
qbPast['Avg'] = qbPast.groupby('Player').rolling(3)['Points'].mean().droplevel(0)
英文:
The problem is index alignment. Don't use reset_index instead drop the zeroth index level then assign the values to a column
qbPast['Avg'] = qbPast.groupby('Player').rolling(3)['Points'].mean().droplevel(0)
答案2
得分: 1
将.reset_index(drop=True)更改为.reset_index(0, drop=True)，以使玩家的索引不混在一起。
英文:
You have to change the .reset_index(drop=True) into .reset_index(0, drop=True) so it is not mixing the players indices together.
答案3
得分: 0
使用groupby - transform是避免索引问题的一个不错的方法：
qbPast['avg'] = (qbPast.
sort_values('Week').
groupby('Player')['Points'].
transform(lambda g: g.rolling(3).mean()))
英文:
A nice way of avoiding the index problem is using groupby - transform:
qbPast['avg'] = (qbPast.
sort_values('Week').
groupby('Player')['Points'].
transform(lambda g: g.rolling(3).mean()))
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论