Pandas - 只能将大小为 1 的数组转换为 Python 标量

Question

我有两个数据框：

df_melt ：

    MatchID GameWeek        Date                      Team  Home               AgainstTeam
0     46605        1  2019-08-09                 Liverpool  Home              Norwich City
1     46605        1  2019-08-09              Norwich City  Away                 Liverpool
2     46606        1  2019-08-10           AFC Bournemouth  Home          Sheffield United
3     46606        1  2019-08-10          Sheffield United  Away           AFC Bournemouth
4     46607        1  2019-08-10                   Burnley  Home               Southampton
..      ...      ...         ...                       ...   ...                       ...
533   46871       27  2020-02-23                   Watford  Away         Manchester United
534   46872       27  2020-02-22          Sheffield United  Home  Brighton and Hove Albion
535   46872       27  2020-02-22  Brighton and Hove Albion  Away          Sheffield United
536   46873       27  2020-02-22               Southampton  Home               Aston Villa
537   46873       27  2020-02-22               Aston Villa  Away               Southampton

并且，对于球员比赛， df_pm ：

                                       Player  GameWeek  Minutes  ... CloseShotCreated TotalShotCreated  HeadersCreated
PlayerMatchesDetailID                                             ...                                                  
1                                     Alisson         1       90  ...                0                0               0
2                             Virgil van Dijk         1       90  ...                0                0               0
3                                Joseph Gomez         1       90  ...                0                1               0
4                            Andrew Robertson         1       90  ...                0                1               0
5                      Trent Alexander-Arnold         1       90  ...                3                3               1
...                                       ...       ...      ...  ...              ...              ...             ...
15053                             Matty James        22        0  ...                0                0               0
15054                             Matty James        23        0  ...                0                0               0
15055                             Matty James        24        0  ...                0                0               0
15056                             Matty James        25        0  ...                0                0               0
15057                             Matty James        26        0  ...                0                0               0

现在，我正在尝试遍历df_pm并在df_melt给定某些条件下查找项目，如下所示：

#Instantiate an empty list
match_ids = []
home_away = []
dates = []

#For each row in the player matches dataframe...
for row in df_pm.itertuples():
    #Look up the match id from the team matches dataframe
    team = row.ForTeam
    againstteam = row.AgainstTeam
    gameweek = row.GameWeek

    match_id = df_melt.loc[(df_melt['GameWeek']==gameweek)
                          &(df_melt['Team']==team)
                          &(df_melt['AgainstTeam']==againstteam),
                          'MatchID'].item()

    date = df_melt.loc[(df_melt['GameWeek']==gameweek)
                          &(df_melt['Team']==team)
                          &(df_melt['AgainstTeam']==againstteam),
                          'Date'].item()

    home = df_melt.loc[(df_melt['GameWeek']==gameweek)
                          &(df_melt['Team']==team)
                          &(df_melt['AgainstTeam']==againstteam),
                          'Home'].item()

    #Add it to the list
    match_ids.append(match_id)
    home_away.append(home)
    dates.append(date)

但是对于所有迭代，即使我打印“团队”、“对抗团队”和“游戏周”，我也会收到以下错误：

Traceback (most recent call last):
  File "tableau_data_generation.py", line 155, in <module>
    'MatchID'].item()
  File "/Users/me/anaconda2/envs/data_science/lib/python3.7/site-packages/pandas/core/base.py", line 652, in item
    return self.values.item()
ValueError: can only convert an array of size 1 to a Python scalar

...表明该项目不存在。

但是当我打印完整的 dataframe df_melt ，如下所示：

with pd.option_context('display.max_rows', None, 'display.max_columns', None):  # more options can be specified also
    print(df_melt, df_melt.shape)

我得到(538, 6)并且可以看到所有数据都在那里，没有任何缺陷。

---

当我检查类型时，我看到：

df_melt ：

MatchID        object
GameWeek       object
Date           object
Team           object
Home           object
AgainstTeam    object

df_pm ：

Player                 object
GameWeek                int64
Minutes                 int64
ForTeam                object
AgainstTeam            object
Goals                   int64
ShotsOnTarget           int64
ShotsInBox              int64
CloseShots              int64
TotalShots              int64
Headers                 int64
GoalAssists             int64
ShotOnTargetCreated     int64
ShotInBoxCreated        int64
CloseShotCreated        int64
TotalShotCreated        int64
HeadersCreated          int64

---

所以这里的类型不匹配。

---

如果我在执行迭代之前添加以下代码行：

df_melt['GameWeek'] = pd.to_numeric(df_melt['GameWeek'])

我成功地为df_pm.itertuples()中的第一行打印了数十个“match_id”、“日期”和“家”（在我添加该行之前没有打印），只是在第二行再次中断错误：

ValueError: can only convert an array of size 1 to a Python scalar

---

我该如何解决？

---

注意：这是上面代码之后的内容。

def matchid_lookup(player, date, team, gameweek):
    try:
        try:
            return df_pm.loc[(df_pm['Date']==date)
                        &(df_pm['Player']==player), 'MatchID'].item()
        except:
            return df_pm.loc[(df_pm['Date']==date)
                        &(df_pm['ForTeam']==team), 'MatchID'].iloc[0]
    except:
        return df_pm.loc[(df_pm['GameWeek']==gameweek)
                        &(df_pm['Player']==player), 'MatchID'].item()

#Declare the list as a column in the player matches df
df_pm['MatchID']=match_ids
df_pm['Date']=pd.to_datetime(dates)
df_pm['Home']=home_away
df_pm['Position']=df_pm['Player'].map(pos_lookup)

#Get the match IDs column first in the dataframe
cols = list(df_pm.columns)
new_cols = ['MatchID', 'Date', 'Home','Position'] + cols[:-4]
df_pm = df_pm[new_cols]

#Bring in stats from api table
#First, get key identifiers into the api table to facilitate joining
df_api_stats['Player'] = df_api_stats['PlayerID'].map(player_lookup)
df_api_stats['Team'] = df_api_stats['PlayerID'].map(team_lookup)    
df_api_stats['MatchID'] = df_api_stats.apply(lambda x: matchid_lookup(x['Player'],
                                                                      x['Date'],
                                                                      x['Team'],
                                                                      x['GameWeek']), axis=1)
api_cols = ['Player', 'MatchID', 'BPS', 'MinutesPlayed',
            'CleanSheet', 'Saves', 'NetTransfersIn',
            'SelectedBy', 'Points', 'Price']

df_api_cols = df_api_stats[api_cols]

Answer 1

所以有一些来自df_api_stats的日期不在df_pm中，您可以通过以下方式查看：

print (set(pd.to_datetime(df_api_stats['Date'])) - set(pd.to_datetime(df_pm['Date'])))
{Timestamp('2020-01-29 00:00:00'),
 Timestamp('2020-02-28 00:00:00'),
 Timestamp('2020-02-29 00:00:00'),
 Timestamp('2020-03-01 00:00:00'),
 Timestamp('2020-03-07 00:00:00'),
 Timestamp('2020-03-08 00:00:00'),
 Timestamp('2020-03-09 00:00:00')}

我不确定你想对缺失值做什么，但为了避免方法失败，你可以添加一个 except 并在没有任何可能匹配的情况下返回 nan。

def matchid_lookup(player, date, team, gameweek):
    try:
        try:
            return df_pm.loc[(df_pm['Date']==date)
                        &(df_pm['Player']==player), 'MatchID'].item()
        except:
            return df_pm.loc[(df_pm['Date']==date)
                        &(df_pm['ForTeam']==team), 'MatchID'].iloc[0]
    except:
        try:
            return df_pm.loc[(df_pm['GameWeek']==gameweek)
                            &(df_pm['Player']==player), 'MatchID'].item()
        except:
            return np.nan

注意：就在之前导致问题的for循环之前，不要忘记这样做：

df_melt['GameWeek'] = pd.to_numeric(df_melt['GameWeek'])
df_melt[['Team', 'AgainstTeam']] = df_melt[['Team', 'AgainstTeam']]\
                                          .replace('AFC Bournemouth', 'Bournemouth')

Answer 2

 return self.values.item()
 ValueError: can only convert an array of size 1 to a Python scalar

上面的错误是说你有一个包含多个元素的数组。 为了能够使用.item()，您应该只有一个值，以便它可以从数组转换为标量。