python数据可视化

2020-09-07

pycharts

pyecharts官网

参考官网上的示例我们可以可视化数据，渲染成html文件

注意：如果需要保存为图片，还需要安装selenium, phantomjs 和 pyppeteer 相应插件来处理

from pyecharts.charts import Bar
from pyecharts import options as opts
# V1 版本开始支持链式调用
# 你所看到的格式其实是 `black` 格式化以后的效果
# 可以执行 `pip install black` 下载使用
bar = (
    Bar()
    .add_xaxis(["衬衫", "羊毛衫", "雪纺衫", "裤子", "高跟鞋", "袜子"])
    .add_yaxis("商家A", [5, 20, 36, 10, 75, 90])
    .set_global_opts(title_opts=opts.TitleOpts(title="主标题", subtitle="副标题"))
    # 或者直接使用字典参数
    # .set_global_opts(title_opts={"text": "主标题", "subtext": "副标题"})
)
bar.render()
# 不习惯链式调用的开发者依旧可以单独调用方法
bar = Bar()
bar.add_xaxis(["衬衫", "羊毛衫", "雪纺衫", "裤子", "高跟鞋", "袜子"])
bar.add_yaxis("商家A", [5, 20, 36, 10, 75, 90])
bar.set_global_opts(title_opts=opts.TitleOpts(title="主标题", subtitle="副标题"))
bar.render()

PyGraphviz

PyGraphviz (几何图形可视化工具) 简单入门
Graphviz 是一个几何图形可视化软件，能够将数据的结构信息表示为抽象图形和网络图的方式。
PyGraphviz 主要对数据量大、可任意定制的图形上具有强大优势。

安装：

brew install graphviz # mac
pip install graphviz # linux
conda install graphviz # conda
pip3 install pygraphviz

使用：

神经网络图

神经网络.png

import pygraphviz as pgv
G = pgv.AGraph(directed=True, rankdir="LR")
# 设置层数、每层节点数
network_structure = input("输入神经网络每层节点数，如：[10, 7, 5, 4, 2, 1]\n")
all_nodes = [[f"layer{layer_index}node{node_index}" for node_index in range(layer_nodes_num)] for
             layer_index, layer_nodes_num in enumerate(eval(network_structure))]
# 添加输入层节点
for node_index, node in enumerate(all_nodes[0]):
    G.add_node(node, label="", shape="circle", style="bold", color="#60acfc", pos=f"0,{0.6 * node_index}!")
# 添加输入层文本
G.add_node(f"Input_Layer", shape="none", label=f"Input_Layer({len(all_nodes[0])})",
           pos=f"{0},{(node_index + 1) * 0.6}!")
# 添加中间层、输出层节点
for layer_index, layer in enumerate(all_nodes[1:]):
    color = "#5bc49f" if layer_index == len(all_nodes[1:]) - 1 else "#ff7c7c"
    for node_index, node in enumerate(layer):
        x = 1.5 * (layer_index + 1)
        y = - 0.3 * (len(all_nodes[layer_index + 1]) - len(all_nodes[0])) + 0.6 * node_index
        G.add_node(node, label="", shape="circle", style="bold", color=color, pos=f"{x},{y}!")
    # 添加中间层、输出层文本
    text = f"Output_Layer({len(all_nodes[-1])})" if layer_index == len(all_nodes[1:]) - 1 else f"Layer_{layer_index + 1}({len(layer)})"
    G.add_node(f"Layer_{layer_index + 1})", shape="none", label=text,
               pos=f"{x},{y + 0.4}!")
# 添加线
for layer_index in range(all_nodes.__len__() - 1):
    for start in all_nodes[layer_index]:
        for end in all_nodes[layer_index + 1]:
            G.add_edge(start, end)
# 导出图形
G.layout()
G.draw("神经网络.png")
# 运行代码，输入 [5, 10, 15, 15, 15, 10, 5, 1]

因子相关性

因子相关性图.png

import pygraphviz as pgv
G = pgv.AGraph(directed=True, rankdir="TB")
# 设置节点标签
Root = "道路交通流畅"
negative_1 = "平均延误时间"
negative_2 = "负荷度"
negative_3 = "小区位置"
negative_4 = "相对延误率"
negative_5 = "房屋密度"
negative_6 = "人口密度"
negative_7 = "总延误率"
negative_8 = "排队率"
negative_9 = "行驶时间"
positive_1 = "通行能力"
positive_2 = "公路层级"
positive_3 = "路网结构"
positive_4 = "行驶速度"
positive_5 = "路网长度"
positive_6 = "小区面积"
positive_7 = "内部道路密度"
positive_8 = "路网密度"
# 添加节点
G.add_node(Root, style="filled", shape="box3d", color="#feb64d")
for negative in [eval(_) for _ in dir() if _.startswith("negative")]:
    G.add_node(negative, style="filled", shape="ellipse", color="#CFDBF6")
for positive in [eval(_) for _ in dir() if _.startswith("positive")]:
    G.add_node(positive, style="filled", shape="ellipse", color="#B4E7B7")
# 添加边
G.add_edges_from([[Root, negative_1], [Root, negative_6], [Root, negative_8], [Root, negative_9],
                  [negative_1, negative_2], [negative_1, negative_7], [negative_2, negative_3],
                  [negative_2, negative_7], [negative_3, negative_4], [negative_8, negative_9],
                  [positive_2, negative_5], [positive_3, negative_4], [positive_4, negative_5]],
                 color="#B4DBFF", style="dashed", penwidth=1.5)
G.add_edges_from([[Root, positive_1], [Root, positive_8], [negative_5, negative_4],
                  [negative_6, positive_4], [negative_5, positive_4], [negative_9, positive_5],
                  [positive_1, positive_2], [positive_2, positive_3], [positive_6, positive_5],
                  [positive_7, positive_6], [positive_8, positive_7]],
                 color="#B4E7B7", style="dashed", penwidth=1.5)
# 导出图形
G.layout()
G.draw("因子相关性图.png", prog="dot")

SmartChart

smartchart