計算 DataFrame 的每個部分/切片的單獨平均值

Question

對於當前項目，我計划計算rating_recommend的平均值。 object 被stock_symbol切片作為第一和quarter作為第二切片標准。

目前，output 文件顯示了所有類別的一般平均值，如下所示：

stock_symbol  quarter   rating_recommend 
A             2008Q2    1.270 
A             2008Q3    1.270 
A             2008Q4    1.270 
A             2009Q1    1.270 
A             2009Q2    1.270 
A             2009Q3    1.270

目標是為每個類別獲得單獨的手段：

stock_symbol  quarter   rating_recommend 
A             2008Q2    1.123 
A             2008Q3    1.321 
A             2008Q4    1.674 
A             2009Q1    1.003
A             2009Q2    1.245 
A             2009Q3    1.177

是否有任何聰明的調整來使這項工作？ 相關代碼部分如下所示：

# Datetime conversion
df['date'] = pd.to_datetime(df['date'])
df['quarter'] = df['date'].dt.to_period('Q')

# Definition of the data objects
def get_top_n_bigram(row):
    # Convert quantitative data and remove null values
    df['rating_recommend'] = pd.to_numeric(df['rating_recommend'], errors='coerce')
    return df['rating_recommend'].mean()

# Grouping data and assigning this as a new dataframe
newdf = df.groupby(['stock_symbol', 'quarter']).apply(get_top_n_bigram).to_frame(name = 'rating_recommend')

# Exporting the dataframe to Excel
newdf.to_excel('total_bigrams_pro.xlsx')

最后但並非最不重要的一點是，示例數據如下所示：

[
{"gld_index": "1-0", "stock_symbol": "AMG", "gld_id": "7172", "date": "2013-01-01", "rating_recommend": 0, "rating_outlook": 1, "rating_ceo": 1, "scr_avg": 1.0, "scr_balance": 1.0, "scr_values": 1.0, "scr_opportunities": 1.0, "scr_benefits": 1.0, "scr_management": 1.0},
{"gld_index": "1-2", "stock_symbol": "AMG", "gld_id": "7172", "date": "2011-09-15", "rating_recommend": 2, "rating_outlook": null, "rating_ceo": 2, "scr_avg": 4.0, "scr_balance": 5.0, "scr_values": null, "scr_opportunities": 4.0, "scr_benefits": 5.0, "scr_management": 4.5},
{"gld_index": "1-0", "stock_symbol": "MMM", "gld_id": "446", "date": "2017-05-14", "rating_recommend": 2, "rating_outlook": 1, "rating_ceo": 2, "scr_avg": 4.0, "scr_balance": 4.0, "scr_values": 5.0, "scr_opportunities": 3.0, "scr_benefits": 3.0, "scr_management": 4.0}
]

Answer 1

我想它應該工作：

newdf = df.groupby(['stock_symbol', 'quarter']).mean()

Answer 2

您可以一次轉換rating_recommend而不是循環內的每一行：

# Datetime conversion
df['date'] = pd.to_datetime(df['date'])
df['quarter'] = df['date'].dt.to_period('Q')

# Convert quantitative data and remove null values
df['rating_recommend'] = pd.to_numeric(df['rating_recommend'], errors='coerce')

# Grouping data and assigning this as a new dataframe
newdf = df.groupby(['stock_symbol', 'quarter'])['rating_recommend'].mean().reset_index()

# Exporting the dataframe to Excel
newdf.to_excel('total_bigrams_pro.xlsx')