pd.sort_values沒有做它應該

Question

我有我一直在使用 df = pd.read_csv("af.csv")

CSV文件看起來像這樣（預覽）已經導入的CSV文件：

"match_id","start_time","win","leaguename","opposing_team","team","min" 
2992096687,1486840800,True,"CaptainsDraft",3729377,2642171,1453382256 
2992217489,1486845476,true,"Captains Draft",3729377,2642171,1453382256 
2994454005,1486926905,false,"Captains Draft",2586976,2642171,1453382256 
2659805546,1474478411,false,"BTSSeries",55,2642171,1454281287 
2659879628,1474481141,false,"BTSSeries",55,2642171,1454281287 
2661783205,1474563571,false,"BTSSeries",2537636,2642171,1454281287 
2661875544,1474566865,false,"BTSSeries",2537636,2642171,1454281287 
2662027296,1474573160,true,"BTSSeries",59,2642171,1454281287 
2758086417,1478352060,true,"ESLManila16",2163,2642171,1454692269 
2758241073,1478355547,true,"ESLManila16",2163,2642171,1454692269 
2747710178,1477941012,false,"ESLFrankfurt16",2850016,2642171,1459782261 
2747808587,1477945318,true,"ESLFrankfurt16",2850016,2642171,1459782261 
2747861268,1477947994,true,"ESLFrankfurt16",2850016,2642171,1459782261 

現在我」 m試圖做的是保持聯賽的第一場比賽，然後是勝利數（True is a win，False is a loss）全部匹配對該聯賽然後按start_time對其進行排序

我有下面的代碼來做到這一點：

df1 = df.groupby(['leaguename', 'team']).sum().reset_index() 
df1 = df1[['win','leaguename','team']] 

df2 = df.sort_values("start_time").groupby("leaguename", as_index=False).first() 
df2 = df2[['leaguename', 'start_time']] 

output = pd.merge(df1, df2, 'inner', on = 'leaguename') 

輸出返回的混亂與無序START_TIME：

,win,leaguename,team,start_time 
0,5.0,ASUSROGSeason6,2642171,1478022101 
1,6.0,CaptainsDraft,2642171,1486840800 
2,3.0,Dota2Asia17,2642171,1486130597 
3,2.0,DotaPitSeason5,2642171,1476903919 
4,5.0,ESLFrankfurt16,2642171,1477941012 
5,2.0,ESLManila16,2642171,1478352060 
6,6.0,GlobalGrandMasters,2642171,1466176095 
7,4.0,NanyangChampionshipsSeason2,2642171,1464178206 

所需的輸出：

,win,leaguename,team,start_time 
0,4.0,NanyangChampionshipsSeason2,2642171,1464178206 
1,6.0,GlobalGrandMasters,2642171,1466176095 
2,2.0,DotaPitSeason5,2642171,1476903919 
3,5.0,ESLFrankfurt16,2642171,1477941012 
4,5.0,ASUSROGSeason6,2642171,1478022101 
5,2.0,ESLManila16,2642171,1478352060 
6,3.0,Dota2Asia17,2642171,1486130597 
7,6.0,CaptainsDraft,2642171,1486840800 

我如何能實現所需的輸出？

Answer 1

我想你需要通過DataFrame.sort_values列start_time與DataFrame.reset_index和默認獨特的單調指標參數drop=True：

output = output.sort_values('start_time').reset_index(drop=True) 
#data by output sample 
print (output) 
    win     leaguename  team start_time 
0 4.0 NanyangChampionshipsSeason2 2642171 1464178206 
1 6.0   GlobalGrandMasters 2642171 1466176095 
2 2.0    DotaPitSeason5 2642171 1476903919 
3 5.0    ESLFrankfurt16 2642171 1477941012 
4 5.0    ASUSROGSeason6 2642171 1478022101 
5 2.0     ESLManila16 2642171 1478352060 
6 3.0     Dota2Asia17 2642171 1486130597 
7 6.0    CaptainsDraft 2642171 1486840800 

另一種解決方案是增加sort=False既groupby：

df1 = df.groupby(['leaguename', 'team'], sort=False).sum().reset_index() 
df1 = df1[['win','leaguename','team']] 

df2 = df.sort_values("start_time").groupby("leaguename", as_index=False, sort=False).first() 
df2 = df2[['leaguename', 'start_time']] 

output = pd.merge(df1, df2, on = 'leaguename') 
#data by input sample 
print (output) 
    win  leaguename  team start_time 
0 2.0 Captains Draft 2642171 1486840800 
1 1.0  BTSSeries 2642171 1474478411 
2 2.0  ESLManila16 2642171 1478352060 
3 2.0 ESLFrankfurt16 2642171 1477941012