基于Pandas DataFrame的关键路径网络流构建、时长计算及可视化问询

一、计算所有路径的总时长 #

将活动依赖关系转换为有向图结构，通过深度优先搜索（DFS）遍历所有从起始节点（无前置活动）到终点G的路径，并计算每条路径的总时长。

实现代码 #

import pandas as pd
from collections import defaultdict

# 初始化原始数据
Activity = ['A','B','C','D','E','F','G']
Predecessor = [None, None, None, 'A', 'C', 'A', ['B','D','E']]
Durations = [2,6,4,3,5,4,2]
df = pd.DataFrame(zip(Activity, Predecessor, Durations), 
                  columns = ['Activity','Predecessor','Durations'])

# 构建邻接表：key为前置活动，value为(后续活动, 后续活动时长)
adj = defaultdict(list)
duration_map = {row['Activity']: row['Durations'] for _, row in df.iterrows()}

for _, row in df.iterrows():
    act = row['Activity']
    preds = row['Predecessor']
    if preds is None:
        continue
    # 处理单个或多个前置活动的情况
    if isinstance(preds, list):
        for p in preds:
            adj[p].append((act, duration_map[act]))
    else:
        adj[preds].append((act, duration_map[act]))

# 确定起始节点（无前置活动）和目标节点
start_nodes = df[df['Predecessor'].isna()]['Activity'].tolist()
target_node = 'G'

# DFS遍历所有路径并计算总时长
all_paths = []
def dfs(current, path, total_duration):
    if current == target_node:
        all_paths.append((path + [current], total_duration))
        return
    # 遍历当前节点的所有后续节点
    for neighbor, dur in adj.get(current, []):
        dfs(neighbor, path + [current], total_duration + dur)

# 对每个起始节点执行DFS
for start in start_nodes:
    dfs(start, [], duration_map[start])

# 整理结果为DataFrame
result_df = pd.DataFrame(all_paths, columns=['路径', '总时长'])
print(result_df)

输出结果 #

路径  总时长
0  [A, D, G]    7
1     [B, G]    8
2  [C, E, G]   11

二、网络流可视化 #

方案1：NetworkX + Matplotlib（静态可视化） #

使用NetworkX构建有向图，结合Matplotlib生成静态网络图，清晰展示活动依赖关系及时长。

实现代码

import networkx as nx
import matplotlib.pyplot as plt

# 构建有向图
G = nx.DiGraph()
G.add_nodes_from(df['Activity'])

# 添加边及时长属性
for _, row in df.iterrows():
    act = row['Activity']
    preds = row['Predecessor']
    if preds is None:
        continue
    if isinstance(preds, list):
        for p in preds:
            G.add_edge(p, act, weight=row['Durations'])
    else:
        G.add_edge(preds, act, weight=row['Durations'])

# 设置固定布局
pos = nx.spring_layout(G, seed=42)

# 绘制节点、边及标签
nx.draw_networkx_nodes(G, pos, node_size=1500, node_color='lightblue')
nx.draw_networkx_edges(G, pos, arrowstyle='->', arrowsize=20)
nx.draw_networkx_labels(G, pos, font_size=12, font_weight='bold')
edge_labels = nx.get_edge_attributes(G, 'weight')
nx.draw_networkx_edge_labels(G, pos, edge_labels=edge_labels, font_size=10)

# 调整布局并展示
plt.tight_layout()
plt.axis('off')
plt.show()

方案2：PyVis（交互式可视化） #

PyVis可生成交互式HTML网络图，支持拖拽节点、缩放视图，适合复杂流程的可视化。

实现代码

from pyvis.network import Network

# 初始化交互式有向图
net = Network(directed=True, height='600px', width='100%', bgcolor='#222222', font_color='white')

# 添加节点和边（标注时长）
for _, row in df.iterrows():
    act = row['Activity']
    net.add_node(act, label=act, size=20)
    preds = row['Predecessor']
    if preds is None:
        continue
    if isinstance(preds, list):
        for p in preds:
            net.add_edge(p, act, label=str(row['Durations']), width=2)
    else:
        net.add_edge(preds, act, label=str(row['Durations']), width=2)

# 生成HTML文件
net.show('activity_network.html')

运行后会生成activity_network.html，打开即可查看交互式网络图。

内容的提问来源于stack exchange，提问作者Mr_Rakhman