python - 在简单的熊猫图上添加汇总表时，matplotlib 的 savefig 很慢

Question

我一直在使用 matplotlib 以总体可接受的处理时间在自动 pdf 报告中呈现各种图表。但是，当我在具有两个轴的图表下方添加汇总表时，将图形保存到文件时性能会显着下降。

因此，下面是重现此图表的代码：

import matplotlib
import matplotlib.pyplot as plt
import pandas as pd
import time

data = [
    ['2018-01-01', 50.472, 52.2, 45.1305555555556, 50.399,
     0.983086079622108, 1.00],
    ['2018-02-01', 70.907, 77.5, 69.8916666666667, 79.137,
     1.01918252884086, 0.98],
    ['2018-03-01', 86.148, 124.4, 106.855555555556, 116.661,
     0.920305088314199, 0.89],
    ['2018-04-01', 150.314, 152.5, 143.0, 147.185, 0.877857009149397,
     None],
    ['2018-05-01', 162.63, 191.6, 163.786111111111, 173.851,
     0.834110398117208, 0.86],
    ['2018-06-01', 206.60205, 211.6, 208.858333333333, 214.27,
     0.846595650655765, 0.91],
    ['2018-07-01', 97.374, 219.0, 228.866666666667, 232.978,
     0.794293255390368, 0.71],
    ['2018-08-01', 194.064, 186.5, 187.916666666667, 196.659,
     0.843036593263637, 0.81],
    ['2018-09-01', 154.48, 139.1, 150.969444444444, 157.047,
     0.866042591545732, 0.82],
    ['2018-10-01', 85.029, 89.6, 81.4833333333333, 89.923,
     0.890580752916001, 0.90],
    ['2018-11-01', 46.389, 56.7, 45.0388888888889, 49.594,
     1.3243466016473, 0.98],
    ['2018-12-01', 51.132, 46.3, 47.6777777777778, 50.513,
     0.904260369658414, 1.08]]
df = pd.DataFrame(data, columns=['day', 'value1',
                                 'value2', 'value3',
                                 'value4',
                                 'value5', 'value6'])


def get_color(column):
    # some logic to retrieve a hex color
    return '#CCCCCC'


matplotlib.use('agg')
init = time.time()
fig = plt.figure()
ax = plt.subplot(111)
ax2 = ax.twinx()
df.set_index(df['day'])
grouped_columns = ['value1', 'value2', 'value3', 'value4']
grouped_color_list = [get_color(series) for series in grouped_columns]
df[grouped_columns].plot.bar(rot=0, ax=ax, color=grouped_color_list)
for column in ['value5', 'value6']:
    ax2.plot(df[column], label=column, color=get_color(column))
ax.xaxis.set_ticks([])
values_df = df.T
table = plt.table(cellText=values_df.values,
                  rowLabels=df.columns,
                  colLabels=df.index,
                  cellLoc='center', rowLoc='center',
                  loc='bottom')
plotting_time = time.time() - init
print('Chart plot took {:.2f}s'.format(plotting_time))
fig.savefig('output.png', bbox_inches='tight', format='png', dpi=100)
saving_time = time.time() - init
print('savefig took {:.2f}s'.format(saving_time))

这是控制台输出（带表格）：

Chart plot took 0.17s
savefig took 4.48s

这是控制台输出（没有表格）：

Chart plot took 0.13s
savefig took 0.62s

有没有我没有正确使用绘图功能的地方？

标签： pythonpandasmatplotlib