import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
# 设置风格，seaborn有5种基本风格，context表示环境
sns.set(style="white", context="notebook")
# 处理中文问题
sns.set_style('whitegrid', {'font.sans-serif':['simhei', 'Arial']})
data = pd.read_csv('D:\myfiles\study\python\analyse\数据团\城市数据团_数据分析师_体验课_课程资料\数据资料\地市级党委书记数据库（2000-10）.csv', encoding='gbk')
data_gender2 = data[['省级政区名称','性别']]
data_gender2_re = data_gender2[data_gender2['性别'].notnull()]
pt = pd.crosstab(data_gender2_re['省级政区名称'], data_gender2_re['性别'])
pt['女性占比'] = pt['女'] / (pt['女'] + pt['男'])
pt2 = pt.sort_values(by = ['女性占比'], ascending = False)
fig = plt.figure(figsize=(10,4))
x = ['辽宁','山西','吉林','陕西','四川','安徽','江西','山东','福建','云南']
pt2['女性占比百分数'] = pt2['女性占比'] *100
y = pt2[:10]['女性占比百分数'].tolist()
# 设定x，y值
sns.barplot(x, y, palette="BuPu_r")
plt.title('女性占比百分数')
plt.ylabel('Per %')
# 数据可视化：柱状图
sns.despine(bottom=True)
plt.show()
