Python + Pandas + 数据可视化：如何获取每行的百分比并可视化分类数据？

Question

我正在对贷款预测数据集（Pandas 数据框）进行探索性数据分析。此数据框有两列：Property_Area，其值为三种类型 - Rural、Urban、Semiurban。另一列是 Loan_Status 明智值有两种类型 - Y、N。我想绘制这样的图表：沿 X 轴应该有 Property_Area，并且，对于每种类型的 3 个区域，我想显示接受的贷款百分比或沿 Y 轴拒绝。该怎么做？

这是我的数据示例：

data = pd.DataFrame({'Loan_Status':['N','Y','Y','Y','Y','N','N','Y','N','Y','N'], 
       'Property_Area': ['Rural', 'Urban','Urban','Urban','Urban','Urban',
       'Semiurban','Urban','Semiurban','Rural','Semiurban']})

我试过这个：

status = data['Loan_Status']
index = data['Property_Area']
df = pd.DataFrame({'Loan Status' : status}, index=index)
ax = df.plot.bar(rot=0)

data is the dataframe for the original dataset

输出：

编辑： 我能够做我想做的事，但为此我不得不写一段很长的代码：

new_data = data[['Property_Area', 'Loan_Status']].copy()
count_rural_y = new_data[(new_data.Property_Area == 'Rural') & (data.Loan_Status == 'Y') ].count()
count_rural = new_data[(new_data.Property_Area == 'Rural')].count()
#print(count_rural[0])
#print(count_rural_y[0])
rural_y_percent = (count_rural_y[0]/count_rural[0])*100
#print(rural_y_percent)

#print("-"*50)

count_urban_y = new_data[(new_data.Property_Area == 'Urban') & (data.Loan_Status == 'Y') ].count()
count_urban = new_data[(new_data.Property_Area == 'Urban')].count()
#print(count_urban[0])
#print(count_urban_y[0])
urban_y_percent = (count_urban_y[0]/count_urban[0])*100
#print(urban_y_percent)

#print("-"*50)

count_semiurban_y = new_data[(new_data.Property_Area == 'Semiurban') & (data.Loan_Status == 'Y') ].count()
count_semiurban = new_data[(new_data.Property_Area == 'Semiurban')].count()
#print(count_semiurban[0])
#print(count_semiurban_y[0])
semiurban_y_percent = (count_semiurban_y[0]/count_semiurban[0])*100
#print(semiurban_y_percent)

#print("-"*50)

objects = ('Rural', 'Urban', 'Semiurban')
y_pos = np.arange(len(objects))
performance = [rural_y_percent,urban_y_percent,semiurban_y_percent]
plt.bar(y_pos, performance, align='center', alpha=0.5)
plt.xticks(y_pos, objects)
plt.ylabel('Loan Approval Percentage')
plt.title('Area Wise Loan Approval Percentage')

plt.show()

输出：

如果可能的话，您能否建议我一个更简单的方法？

Answer 1

Pandas Crosstabs 和 normalize 会让这一切变得简单

在 pandas 数据框中获取 2+ 列并获取 每一行 百分比的简单方法是使用 pandas crosstab 函数和 normalize = 'index'

---

交叉表函数的查找方式如下：

# Crosstab with "normalize = 'index'". 
df_percent = pd.crosstab(data.Property_Area,data.Loan_Status,
                         normalize = 'index').rename_axis(None)

# Multiply all percentages by 100 for graphing. 
df_percent *= 100

这将输出df_percent，如下所示：

Loan_Status          N          Y
Rural        50.000000  50.000000
Semiurban    66.666667  33.333333
Urban        16.666667  83.333333

然后您可以非常轻松地将其绘制到您的条形图中：

# Plot only approvals as bar graph. 
plt.bar(df_percent.index, df_percent.Y, align='center', alpha=0.5)
plt.ylabel('Loan Approval Percentage')
plt.title('Area Wise Loan Approval Percentage')

plt.show()

并得到结果图表：

Here you can see the code working in google colab

---

这是我为此答案生成的示例数据框：

data = pd.DataFrame({'Loan_Status':['N','Y','Y','Y','Y','N','N','Y','N','Y','Y'
   ], 'Property_Area': ['Rural', 'Urban','Urban','Urban','Urban','Urban',
   'Semiurban','Urban','Semiurban','Rural','Semiurban']})

创建此示例数据框：

   Loan_Status Property_Area
0            N         Rural
1            Y         Urban
2            Y         Urban
3            Y         Urban
4            Y         Urban
5            N         Urban
6            N     Semiurban
7            Y         Urban
8            N     Semiurban
9            Y         Rural
10           Y     Semiurban