按小时重新采样 Pandas DataFrame 并使用 Plotly 绘制堆积条形图

Question

我有一个如下的熊猫数据框

MAC Address	ts	Parameter1	Parameter2
af3d116c	2021-05-05 21:58:45	20	50
bffe479a	2021-05-05 21:58:48	22	52
c3a8fe37	2021-05-05 21:58:52	21	53
af3d116c	2021-05-05 21:58:58	27	50
bffe479a	2021-05-05 21:59:16	23	51
c3a8fe37	2021-05-05 21:59:50	28	52
af3d116c	2021-05-05 22:08:32	30	49
af3d116c	2021-05-05 22:16:30	27	55
bffe479a	2021-05-05 22:31:37	20	53
c3a8fe37	2021-05-05 22:52:49	32	52
af3d116c	2021-05-05 23:22:02	41	58
bffe479a	2021-05-05 23:44:31	37	62
bffe479a	2021-05-05 23:45:12	29	58
bffe479a	2021-05-05 23:49:28	34	41
c3a8fe37	2021-05-05 23:52:47	47	56

我想重新采样数据框，最后绘制堆叠条形图（最好使用 plotly），表示每小时记录的总行数，并根据 MA​​C 地址进行颜色编码。

下面是我希望它如何可视化的表示。 （抱歉，它没有使用上面列出的数据，但表明我希望它是怎样的。每个条形代表一个小时，例如：22:00 到 23:00，用代表 MAC 地址的颜色分隔。）

Answer 1

您可以使用 pd.Grouper(key='ts', freq='1h') 进行 groupby 以按小时“重新采样”数据帧。 size 将为您提供 MAC 地址的频率计数：

import pandas as pd
import plotly.express as px

data = {'MAC Address': {1: 'af3d116c', 2: 'bffe479a', 3: 'c3a8fe37', 4: 'af3d116c', 5: 'bffe479a', 6: 'c3a8fe37', 7: 'af3d116c', 8: 'af3d116c', 9: 'bffe479a', 10: 'c3a8fe37', 11: 'af3d116c', 12: 'bffe479a', 13: 'bffe479a', 14: 'bffe479a', 15: 'c3a8fe37'}, 'ts': {1: '2021-05-05 21:58:45', 2: '2021-05-05 21:58:48', 3: '2021-05-05 21:58:52', 4: '2021-05-05 21:58:58', 5: '2021-05-05 21:59:16', 6: '2021-05-05 21:59:50', 7: '2021-05-05 22:08:32', 8: '2021-05-05 22:16:30', 9: '2021-05-05 22:31:37', 10: '2021-05-05 22:52:49', 11: '2021-05-05 23:22:02', 12: '2021-05-05 23:44:31', 13: '2021-05-05 23:45:12', 14: '2021-05-05 23:49:28', 15: '2021-05-05 23:52:47'}, 'Parameter1': {1: 20, 2: 22, 3: 21, 4: 27, 5: 23, 6: 28, 7: 30, 8: 27, 9: 20, 10: 32, 11: 41, 12: 37, 13: 29, 14: 34, 15: 47}, 'Parameter2': {1: 50, 2: 52, 3: 53, 4: 50, 5: 51, 6: 52, 7: 49, 8: 55, 9: 53, 10: 52, 11: 58, 12: 62, 13: 58, 14: 41, 15: 56}}
df = pd.DataFrame(data)
df['ts'] = pd.to_datetime(df['ts'])

plot_df = df.groupby([pd.Grouper(key='ts', freq='1h'), 'MAC Address']).size().reset_index().rename(columns={0: "count"})

这将导致：

	ts	MAC Address	count
0	2021-05-05 21:00:00	af3d116c	2
1	2021-05-05 21:00:00	bffe479a	2
2	2021-05-05 21:00:00	c3a8fe37	2
3	2021-05-05 22:00:00	af3d116c	2
4	2021-05-05 22:00:00	bffe479a	1
5	2021-05-05 22:00:00	c3a8fe37	1
6	2021-05-05 23:00:00	af3d116c	1
7	2021-05-05 23:00:00	bffe479a	3
8	2021-05-05 23:00:00	c3a8fe37	1

然后，您可以根据需要进行绘制。例如：

fig = px.bar(plot_df, x="ts", y="count", color="MAC Address", title="MAC Addresses per hour")
fig.show()