根據不同數據框中的匹配值，將摘要列添加到pandas數據框中

Question

我有一個表示項目成員和項目開始日期的DataFrame，以及表示出生日期的第二個DataFrame。 我正在嘗試根據每個項目的開頭添加一些列，表明某些年齡組的總人數。

print(projects)
           Start  John  Bob  Gladys
Project                               
A     2014-01-08     1    0       0
B     2016-08-09     0    1       1
C     2018-02-06     0    1       0

print(birthdays)
             birth
name              
John    1983-04-06
Gladys  1969-08-02
Bob     1946-11-03

我已經考慮過使用.apply()或.iterrows()方法，但是我很難把頭包裹在哪里開始。 真正的DataFrame有更多的列和行，所以我需要避免需要按名稱調用任何列。

這就是我想要完成的事情：

              Start  John  Bob  Gladys  25-34  35-45  46-55  56+
Project                                                         
A        2014-01-08     1    0       0      1      0      0    0
B        2016-08-09     0    1       1      0      0      1    1
C        2018-02-06     0    1       0      0      0      0    1

有關從哪里開始的任何建議？

Answer 1

這是使用melt ， cut ， crosstab和merge一種方式：

##unpivot the dataframe keeping Start as index and keep only value ==1
melt_=projects.melt('Start').query('value==1') 
#map the variable column from above with our birthday` dataframe squeezed into a series. 
#Once mapped, subtract with the `Start` column to find out year difference.
Age=(melt_.Start-melt_.variable.map(birthdays.squeeze())).dt.days//365
#using pd.cut , we cut the Age into bins and assign our labels.
bins=[25,35,45,55,np.inf]
labels=['25-34','35-45','46-55','56+']
melt_=melt_.assign(Age=pd.cut(Age,bins,labels=labels))

最后使用pd.crosstab(melt_.Start,melt_.Age)的crosstab輸出與原始df合並：

projects.merge(pd.crosstab(melt_.Start,melt_.Age).reindex(columns=labels,fill_value=0)
                              ,left_on='Start',right_index=True)

---

              Start  John  Bob  Gladys  25-34  35-45  46-55  56+
Project                                                        
A       2014-01-08     1    0       0      1      0      0    0
B       2016-08-09     0    1       1      0      0      1    1
C       2018-02-06     0    1       0      0      0      0    1

注意 ：預期的輸出是錯誤的，因為Bob是老年人，根據項目日期是56+ 。

Answer 2

這應該會給你想要的結果：

import pandas as pd
from datetime import datetime

#create the Projects dataframe
d1={'Projects': ['A', 'B','C'], 'Start': ['2014-01-08', '2016-08-09','2018-02-06'],'John':[1,0,0],'Bob':[0,1,1],'Gladys':[0,1,0]}
projects=pd.DataFrame(data=d1)
projects.set_index('Projects',inplace=True)

#create the Birthdays dataframe
d2={'name':['John','Bob','Gladys'],'birth':['1983-04-06','1946-11-03','1969-08-02']}
birthdays=pd.DataFrame(data=d2)
birthdays.set_index('name', inplace=True)

# Create the age slab dataframe with dummy values
d3={'Projects': ['A','B','C'], '25-34':[0,0,0], '35-45':[0,0,0], '46-55':[0,0,0], '56+':[0,0,0]}
age_slabs=pd.DataFrame(data=d3)
age_slabs.set_index('Projects',inplace=True)

#concatenate projects and age_slabs dataframe
concat=pd.concat([projects, age_slabs], axis=1)
concat=concat.copy()

pd.options.mode.chained_assignment = None

date_format = "%Y-%m-%d"

for i in range(len(concat)):

    val_John=((concat['John'][i]*(datetime.strptime(concat['Start'][i], date_format)-datetime.strptime(birthdays['birth'].John, date_format))).days)/365.
    val_Bob=((concat['Bob'][i]*(datetime.strptime(concat['Start'][i], date_format)-datetime.strptime(birthdays['birth'].Bob, date_format))).days)/365.
    val_Gladys=((concat['Gladys'][i]*(datetime.strptime(concat['Start'][i], date_format)-datetime.strptime(birthdays['birth'].Gladys, date_format))).days)/365.

    # check for conditions and set 1 or 0, and add up
    concat['25-34'][i] = 1 if ((val_John<=34) & (val_John>=25)) else 0 + 1 if ((val_Bob<=34) & (val_Bob>=25)) else 0 + 1 if ((val_Gladys<=34) & (val_Gladys>=25)) else 0
    concat['35-45'][i] = 1 if ((val_John<=45) & (val_John>=35)) else 0 + 1 if ((val_Bob<=45) & (val_Bob>=35)) else 0 + 1 if ((val_Gladys<=45) & (val_Gladys>=35)) else 0
    concat['46-55'][i] = 1 if ((val_John<=55) & (val_John>=46)) else 0 + 1 if ((val_Bob<=55) & (val_Bob>=46)) else 0 + 1 if ((val_Gladys<=55) & (val_Gladys>=46)) else 0
    concat['56+'][i] = 1 if ((val_John>=56)) else 0 + 1 if ((val_Bob>=56)) else 0 + 1 if ((val_Gladys>=56)) else 0

print (concat)

輸出：

               Start  John  Bob  Gladys  25-34  35-45  46-55  56+
Projects                                                         
A         2014-01-08     1    0       0      1      0      0    0
B         2016-08-09     0    1       1      0      0      1    1
C         2018-02-06     0    1       0      0      0      0    1