数据处理实例：2012年联邦选举委员会数据库

首先  数据下载网站：

数据集-阿里云天池

下载下来数据后，需要对其进行合并，合并方法：

data_01 = pd.read_csv('data_01.csv')
data_02 = pd.read_csv('data_02.csv')
data_03 = pd.read_csv('data_03.csv')
data = pd.concat([data_01,data_02,data_03])

合并数据后，对数据进行处理。

数据预处理

print(data.describe())
print(data.info())

运行结果：

 可以看出，info（）函数主要是判断里面是否含有空变量，如果存在的话，则可以用fillna来进行处理。

添加党派关系

temp=data['cand_nm'].unique()
#print(temp)
parties = {'Bachmann, Michelle': 'Republican',
           'Cain, Herman': 'Republican',
           'Gingrich, Newt': 'Republican',
           'Huntsman, Jon': 'Republican',
           'Johnson, Gary Earl': 'Republican',
           'McCotter, Thaddeus G': 'Republican',
           'Obama, Barack': 'Democrat',
           'Paul, Ron': 'Republican',
           'Pawlenty, Timothy': 'Republican',
           'Perry, Rick': 'Republican',
           "Roemer, Charles E. 'Buddy' III": 'Republican',
           'Romney, Mitt': 'Republican',
           'Santorum, Rick': 'Republican'}
data['parties']=data['cand_nm'].map(parties)
print(data['parties'].value_counts())

 unique（）:该函数的作用是得到里面的唯一量

map（）：将人名映射到列表里面。

运行结果：

 根据职业和雇主统计赞助信息

没有什么新东西，但是需要仔细钻研代码

代码为：

occupation_map = {
  'INFORMATION REQUESTED PER BEST EFFORTS':'NOT PROVIDED',
  'INFORMATION REQUESTED':'NOT PROVIDED',
  'SELF' : 'SELF-EMPLOYED',
  'SELF EMPLOYED' : 'SELF-EMPLOYED',
  'C.E.O.':'CEO',
  'LAWYER':'ATTORNEY',
}
emp_mapping = {
   'INFORMATION REQUESTED PER BEST EFFORTS' : 'NOT PROVIDED',
   'INFORMATION REQUESTED' : 'NOT PROVIDED',
   'SELF' : 'SELF-EMPLOYED',
   'SELF EMPLOYED' : 'SELF-EMPLOYED',
}

f=lambda g:occupation_map.get(g,g)
data['contbr_occupation']=data['contbr_occupation'].map(f)
f=lambda g:emp_mapping.get(g,g)
data.contbr_employer = data.contbr_employer.map(f)
by_occ=data.pivot_table('contb_receipt_amt',index='contbr_occupation',columns='parties',aggfunc='sum')
over_2=by_occ[by_occ.sum(1)>2000000]
print(over_2)
over_2.plot(kind='barh')
plt.show()

 主要的是添加了映射，其实这个用groupby也是可以的

运行结果

按照每个人捐助的金额进行排序代码：

data=data[data['contb_receipt_amt']>0]
print(data.groupby('cand_nm')['contb_receipt_amt'].sum().sort_values())

 运行结果为：

对出资额分组

先上代码：

data=data[data['contb_receipt_amt']>0]
#print(data.groupby('cand_nm')['contb_receipt_amt'].sum().sort_values())
bins = np.array([0,1,10,100,1000,10000,100000,1000000,10000000])
data_vs = data[data['cand_nm'].isin(['Obama, Barack','Romney, Mitt'])]
labels=pd.cut(data_vs['contb_receipt_amt'],bins)
#print(labels)
print(data_vs.groupby(['cand_nm',labels]).size().unstack(0))

其中，看到最后一句

print(data_vs.groupby(['cand_nm',labels]).size().unstack(0))

size（）的意义是  得到个数，如果将其改成sum（），则得到的是总金额

unstace（0）表示不分开表示两个表格，如果变成  unstack（），则会将其行列反转。

上述代码跑出的结果如下：

 之后，我们算出每个捐赠金额的所占比例，然后根据比例画出图表，代码如下所示：

temp2=data_vs.groupby(['cand_nm',labels]).sum().unstack(0)
temp3=temp2.div(temp2.sum(axis=1),axis=0)
print(temp2.div(temp2.sum(axis=1),axis=0))
temp3.plot(kind='barh',stacked=True)
plt.show()

 在这个代码中，我们可以看到，用到了div（）函数，该函数的主要作用是除法，即给每个变量除以函数的第一个所给值

画图时，指定stacked=True进行堆叠，即可完成百分比堆积图

运行结果为

 完整代码：

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
data_01 = pd.read_csv('data_01.csv')
data_02 = pd.read_csv('data_02.csv')
data_03 = pd.read_csv('data_03.csv')
data = pd.concat([data_01,data_02,data_03])
#print(data.describe())
#print(data.info())
temp=data['cand_nm'].unique()
#print(temp)
parties = {'Bachmann, Michelle': 'Republican',
           'Cain, Herman': 'Republican',
           'Gingrich, Newt': 'Republican',
           'Huntsman, Jon': 'Republican',
           'Johnson, Gary Earl': 'Republican',
           'McCotter, Thaddeus G': 'Republican',
           'Obama, Barack': 'Democrat',
           'Paul, Ron': 'Republican',
           'Pawlenty, Timothy': 'Republican',
           'Perry, Rick': 'Republican',
           "Roemer, Charles E. 'Buddy' III": 'Republican',
           'Romney, Mitt': 'Republican',
           'Santorum, Rick': 'Republican'}
data['parties']=data['cand_nm'].map(parties)
#print(data['parties'].value_counts())
occupation_map = {
  'INFORMATION REQUESTED PER BEST EFFORTS':'NOT PROVIDED',
  'INFORMATION REQUESTED':'NOT PROVIDED',
  'SELF' : 'SELF-EMPLOYED',
  'SELF EMPLOYED' : 'SELF-EMPLOYED',
  'C.E.O.':'CEO',
  'LAWYER':'ATTORNEY',
}
emp_mapping = {
   'INFORMATION REQUESTED PER BEST EFFORTS' : 'NOT PROVIDED',
   'INFORMATION REQUESTED' : 'NOT PROVIDED',
   'SELF' : 'SELF-EMPLOYED',
   'SELF EMPLOYED' : 'SELF-EMPLOYED',
}

f=lambda g:occupation_map.get(g,g)
data['contbr_occupation']=data['contbr_occupation'].map(f)
f=lambda g:emp_mapping.get(g,g)
data.contbr_employer = data.contbr_employer.map(f)
by_occ=data.pivot_table('contb_receipt_amt',index='contbr_occupation',columns='parties',aggfunc='sum')
over_2=by_occ[by_occ.sum(1)>2000000]
#print(over_2)
#over_2.plot(kind='barh')
#plt.show()
data=data[data['contb_receipt_amt']>0]
#print(data.groupby('cand_nm')['contb_receipt_amt'].sum().sort_values())
bins = np.array([0,1,10,100,1000,10000,100000,1000000,10000000])
data_vs = data[data['cand_nm'].isin(['Obama, Barack','Romney, Mitt'])]
labels=pd.cut(data_vs['contb_receipt_amt'],bins)
#print(labels)
temp2=data_vs.groupby(['cand_nm',labels]).sum().unstack(0)
#print(data_vs.groupby(['cand_nm',labels]).sum().unstack(0))
temp3=temp2.div(temp2.sum(axis=1),axis=0)
#print(temp2.div(temp2.sum(axis=1),axis=0))
temp3.plot(kind='barh',stacked=True)
#plt.show()
grouped=data_vs.groupby(['cand_nm','contbr_st'])
totals=grouped.contb_receipt_amt.sum().unstack(0).fillna(0)
totals=totals[totals.sum(1)>100000]
percent=totals.div(totals.sum(1),axis=0)

参考书籍：《利用Python进行数据分析》