根据条件划分大熊猫

Question

我想计算time_spent_ratio（给定customer_id和给定 month的手机与平板电脑所花费的时间，我在手机上将time_spent除以{{1 }}在平板电脑上。

主表

time_spent

输出表

date_month.  customer_id. device_name  time_spent 

01-01-19     2             phone       140
01-01-19     2             tablet      232
01-02-19     2             phone       159
01-02-19     2             tablet      210
01-03-19     2             phone       193
01-03-19     2             tablet      190

Answer 1

在创建比率列之前，您需要先创建数据透视表。 考虑到您的主要DataFrame是df-

df_pivot = pd.pivot_table(df, index=['date_month','customer_id'], columns='device_name', values='time_spent') #Pivoting on required columns(date and customer_id)
df_flattened = pd.DataFrame(df_pivot.to_records()) #Flatten the pivot table to get a datafrmae
df_flattened['time_spent_ratio'] = df_flattened['phone']/df_flattened['tablet'] #Creating the ratio column

Answer 2

解决此问题的一种方法是将除time_spent以外的所有其他列设置为索引，然后将手机的行除以平板电脑的行。

#set all other columns except time_spent as index
res = df.set_index(["device_name","date_month.","customer_id."])

#divide phone rows by tablet rows
res.loc['phone'].div(res.loc['tablet'])


                            time_spent
date_month. customer_id.    
01-01-19       2            0.603448
01-02-19       2            0.757143
01-03-19       2            1.015789