[星期-时间]维度，解析日期绘制散点图，python

Posted 2022-09-10 zhangphil

[星期-时间]维度，解析日期绘制散点图，python

一批数据，数据里面每一行代表一条事件，每一条事件基本上都包含年月日时分，这样以来，时间就可以代表一次事件，把事件在x坐标轴分解时间（时分），y坐标轴分解为星期几，形成统计图表。最多的事件发生星期-时间点，用红色标记。

import re

import matplotlib.pyplot as plt

import dateutil.parser as ps

from collections import Counter

from pprint import pp

import numpy as np
from fuzzywuzzy import fuzz

FILE_PATH = r'数据文件路径'
KEYS = [r'关键词1', r'关键词2', r'关键词3']
threshold = 90


def change_str(date_s):
    date_s = date_s.replace(r'年', '-')
    date_s = date_s.replace(r'月', '-')
    date_s = date_s.replace(r'日', ' ')

    date_s = date_s.replace(r'时', ':')
    date_s = date_s.replace(r'分', '')

    return date_s


def read_file():
    file = open(FILE_PATH, 'r', encoding='UTF-8')

    all_case_time = []

    case_count = 0
    cnt = 0

    for line in file:
        cnt = cnt + 1
        for k in KEYS:
            pr = fuzz.partial_ratio(line, k)
            if pr >= threshold:
                print('-----')
                print(f'第case_count件')
                case_count = case_count + 1

                try:
                    # 正则匹配 xxxx年xx月xx日xx时xx分
                    mat = re.search(r'\\d4\\年\\d1,2\\月\\d1,2\\日\\d1,2\\时\\d1,2\\分', line)
                    t_str = mat.group().replace('\\n', '')  # 去掉正则匹配到但是多余的 \\n 换行符

                    try:
                        t_str = change_str(t_str)
                        all_case_time.append(ps.parse(t_str))  # 日期提取出来，放到数组中
                    except:
                        print('解析日期失败')
                except:
                    t_str = '-解析异常-'

                s = '第number行,相似度ratio,时间:case_time - content'
                ss = s.format(number=cnt, ratio=pr, case_time=t_str, content=line)
                pp(ss)

                break

        # 快速调试
        # if case_count > 20:
        #    break

    file.close()

    return all_case_time


# 把数字0,1,2,3,4,5,6转换为星期*
# 0为星期一，6为星期日，依次类推
def number_to_weekday(number):
    zh = ['一', '二', '三', '四', '五', '六', '日']
    weekday = f'星期zh[number]'
    return weekday


def date_to_points():
    date_times = read_file()

    hours = []
    for i in range(24):
        hours.append((i, i + 1))

    points = []
    for dt in date_times:
        for h in hours:
            if h[0] <= dt.hour < h[1]:
                point = (h[0], dt.weekday(), h, dt)  # weekday()  # 0是星期一，6是星期日
                points.append(point)
                break

    to_chart(points)


def to_chart(points):
    print('points', points)

    xy = []
    for point in points:
        xy.append((point[0], point[1]))

    print('xy', xy)

    # 找出出现次数最多的点数
    c_xy = Counter(xy)
    print('c_xy', c_xy)

    # 再选取前n多的点数
    max_count = c_xy.most_common(50)
    print('max_count', max_count)

    sizes = []
    colors = []
    for i in range(len(max_count)):
        colors.append('gray')
        sizes.append(30)

    idx = 0
    ax_x = []
    ax_y = []
    for p in max_count:
        ax_x.append(p[0][0])
        ax_y.append(p[0][1])
        sizes[idx] = p[1] * 10 + 10

        if idx < 5:  # 前5红色
            colors[idx] = 'red'

        idx = idx + 1

    plt.style.use('_mpl-gallery')

    plt.rcParams['font.sans-serif'] = ['SimHei']  # 中文乱码
    fig, ax = plt.subplots(figsize=(7, 5))
    plt.subplots_adjust(left=0.1, bottom=0.1, right=0.9)  # 调整边距

    ax.scatter(ax_x, ax_y, s=sizes, c=colors, vmin=0, vmax=100)

    x_labels = []
    for i in range(24):
        x_labels.append((i, i + 1))

    y_labels = []
    for i in range(7):
        y_labels.append(number_to_weekday(i))

    plt.xticks(ticks=np.arange(start=0, stop=24, step=1), labels=x_labels, rotation=45, size=8, color='red')
    plt.yticks(ticks=np.linspace(start=0, stop=7, endpoint=False, num=7, dtype=np.int32), labels=y_labels, rotation=10)

    plt.show()


if __name__ == '__main__':
    date_to_points()

输出：