西电-数据可视化-实验六-层次和网络可视化、文本可视化
一、实验内容
现有xx公司一个月内研发部门成员间的邮件往来记录（email_dev_inside.xlsx，课程资料库下载），请根据以下要求完成对该公司的研发部门的可视化分析：
1、对邮件往来数据进行处理，选择合适的图表，展示公司研发部门的成员组织结构，找出每个群体的负责人。
2、对邮件内容数据进行处理，分析研发部门以及每个群体中的主要工作内容。
二、思路分析
(1)进入pyecharts官网修改demo即可
本实验涉及的图表类型如下：
Graph
WordCloud
Tree
(2)代码实现
#!/usr/bin/env python
# @desc: 层次和网络可视化、文本可视化
import pandas as pd
from pyecharts import options as opts
from pyecharts.charts import Graph, Tree, Page, WordCloud
Link = [] # 全局变量 两个结点之间关系 即邮件通信
infos = [] # 发件人-收件人-主题
# ------------- 绘制关系图-------------------
# 获取绘制关系图的数据
def get_data_for_graph(f):
global Link
# 记录所有的结点 即公司的所有员工 员工用邮件的前四位表示
Node = []
# 遍历文件的每一行
for _, elem in f.iterrows():
# 添加结点
Node.append(elem[6][0:4]) # 发件人
Node.append(elem[7][0:4]) # 收件人
infos.append([elem[6][0:4], elem[7][0:4], elem[8]])
# 建立节点之间的关系
Link.append({
"source": elem[6][0:4], "target": elem[7][0:4]})
# 对Node进行去重
Node = set(Node)
my_Node = [{
"name": node, "symbolsize": 1} for node in Node]
return my_Node, Link
def draw_graph(f):
node, link = get_data_for_graph(f)
(
Graph(opts.InitOpts(width="1000px", height="1000px"))
.add("邮件通信关系图", node, link)
.set_global_opts(title_opts=opts.TitleOpts(title="公司邮件通信状况分析"))
.render("公司邮件通信情况图.html")
)
# ----------------绘制公司人员结构Tree---------------------------------------
# draw_graph后可以看出分为三个部门
three_leader = ["1007", "1059", "1068"]
# 找到每个leader的下属-项目经理
def find_manager_for_leader(leaders):
# {"leader1":[mgr1,mgr2,……],"leader2”:[mgr1,mgr2,……],"leader3":[mgr1,mgr2,……]}
relation = {
}
for leader in leaders:
relation[leader] = []
# Graph中与Leader相连的是他的下属managers
# 遍历Link找与Leader相连的非leader结点即就是mgrs
for link in Link:
if link['source'] in leaders and link["target"] not in leaders:
relation[link["source"]].append(link["target"])
if link['source'] not in leaders and link["target"] in leaders:
relation[link["target"]].append(link["source"])
# 对leadr的下属mgrs去重
for leader in leaders:
relation[leader] = list(set(relation[leader]))
relation[leader].sort()
return relation
# 建立公司的人员结构字典
def creat_person_structure():
the_person_struct = {
}
# {"leader1":{mgr1:[e1,e2,……],mgr2:[],……}],"leader2:{mgr1:[],mgr2:[],……},"leader3":{mgr1:[],mgr2:[],……}}
for leader in three_leader:
the_person_struct[leader] = {
}
relation_1 = find_manager_for_leader(three_leader)
for leader, mgrs in relation_1.items():
# 找mgrs的下属employee 与找leader的下属mgrs原理相同
employee = find_manager_for_leader(mgrs)
# 建立公司人员结构
for mgr in mgrs:
employee[mgr] = [item for item in employee[mgr] if item not in three_leader]
the_person_struct[leader][mgr] = employee[mgr]
return the_person_struct
# 获取绘制Tree的数据
def get_data_for_tree():
data_for_tree = [{
"name": "公司员工组织结构", "children": []}]
# 根据公司的人员结构构建Tree的data
the_person_struct = creat_person_structure()
i = 0 # leader的index
for leader, mgrs in the_person_struct.items():
data_for_tree[0]["children"].append({
"name": leader, "children": []})
# [{"name": "员工结构", "children": [{"name":leader1, "children":[]}]}]}]
# [{"name": "员工结构", "children": [{"name":leader1, "children":[{"name":mgr1,children:[{"name":e1}]}]}]}]
j = 0 # manager的index
for mgr, employees in mgrs.items():
data_for_tree[0]["children"][i]["children"].append({
"name": mgr, "children": []})
for employee in employees:
data_for_tree[0]["children"][i]["children"][j]["children"].append({
"name": employee})
j += 1
i += 1
return data_for_tree
def draw_tree():
(
Tree(opts.InitOpts(width="1500px", height="500px"))
.add(
series_name="公司员工组织结构",
data=get_data_for_tree(),
orient="TB", # 上下结构
collapse_interval=0 # 折叠节点间隔，当节点过多时可以解决节点显示过杂间隔
)
.set_global_opts(title_opts=opts.TitleOpts(title="公司组织结构"))
.render("公司人员组织结构树.html")
)
# -------------------绘制邮件主题词云-------------------------------------
# 获取绘制词云的数据 以下边这种形式保存
# {"leader1":{mgr1:word_list,mgr2:[],……},"leader2:{mgr1:[],mgr2:[],……},"leader3":{mgr1:[],mgr2:[],……}}
def get_data_for_cloud():
struct = creat_person_structure()
# {"leader1":{mgr1:[],mgr2:[],……},"leader2:{mgr1:[],mgr2:[],……},"leader3":{mgr1:[],mgr2:[],……}}
for leader, mgrs in struct.items():
for mgr, employees in mgrs.items():
# 创建一个字典：key是邮件的subject，value是出现的频率
my_dict = {
}
# 遍历所有的邮件 查找leader-mgr下的员工的邮件主题
for info in infos:
if info[0] in employees or info[1] in employees:
# 如果字典中没有这个subject，添加一个key为suject,value为1的键值对
if info[2] not in list(my_dict.keys()):
my_dict[info[2]] = 1
# 如果字典中有这个subject，则value+1
else:
my_dict[info[2]] += 1
# 词云数据
word_list = [(subject, str(counts)) for subject, counts in my_dict.items()]
# {"leader1":{mgr1:word_list,mgr2:[],……},"leader2:{mgr1:[],mgr2:[],……},"leader3":{mgr1:[],mgr2:[],……}}
struct[leader][mgr] = word_list
return struct
# 绘制wordcloud
def draw_cloud():
# 组合图标 Page 顺序多图
page = Page()
# 获取绘制词云的数据
data = get_data_for_cloud()
# leader下的mgr负责这个小群体的邮件主题词云
for leader, mgrs in data.items():
for mgr, word_list in mgrs.items():
word_cloud = (
WordCloud()
.add(
series_name=leader+"领导的部门"+mgr+"管理的团队的邮件主题",
data_pair=data[leader][mgr],
word_size_range=[10, 100]
)
.set_global_opts(
title_opts=opts.TitleOpts(
title=leader+"领导的部门"+mgr+"管理的团队的邮件主题"
)
)
)
page.add(word_cloud)
page.render("邮件主题词云.html")
if __name__ == '__main__':
# 读取文件
filename = "email_dev_inside.xlsx"
df = pd.read_excel(filename)
# 绘制公司员工关系图 Graph
draw_graph(df)
# 绘制公司员工关系结构 Tree
draw_tree()
# 绘制某部门某群体的邮件主题词云
draw_cloud()
三、实验结果
1.邮件通信图（圈起来的就是部门负责人）
2.公司人员结构树（未完全展开）
3.邮件内容词云（部分）
