带有字典数据框的时间序列条形图 - Matplotlib

Question

我正在尝试创建一个时间序列图表，其中我有嵌套的数据框。 最初我的数据格式是：

PartId  Event
100     [{EventId: 1, EventTime: "2022-06-19T10:50:14.513"}, {EventId: 2, EventTime: "2022-07-19T10:50:14.513"}]
...

示例数据：

[
  {
    "PartId": 1000,
    "Events": [
      {
        "Id": 25,
        "EventTime": "2022-04-19T10:33:15.807",
      },
      {
        "Id": 2,
        "EventTime": "2022-04-19T10:50:14.513",
      },
      {
        "Id": 2,
        "EventTime": "2022-04-19T10:50:23.007",
      },
      {
        "Id": 25,
        "EventTime": "2022-04-19T10:52:20.457",
      },
      {
        "Id": 11,
        "EventTime": "2022-04-19T12:30:33.79",
      },
      {
        "Id": 16,
        "EventTime": "2022-04-19T12:38:31.167",
      },
      {
        "Id": 21,
        "EventTime": "2022-04-19T13:03:29.953",
      },
      {
        "Id": 21,
        "EventTime": "2022-04-19T13:03:30.403",
      },
      {
        "Id": 15,
        "EventTime": "2022-04-19T13:57:29.753",
      },
      {
        "Id": 27,
        "EventTime": "2022-04-22T10:29:09.857",
      }
    ]
  },
  {
    "PartId": 1001,
    "Events": [
      {
        "Id": 25,
        "EventTime": "2022-04-19T10:33:15.817",
      },
      {
        "Id": 2,
        "EventTime": "2022-04-19T10:55:14.297",
      },
      {
        "Id": 2,
        "EventTime": "2022-04-19T10:55:22.917",
      },
      {
        "Id": 25,
        "EventTime": "2022-04-19T10:56:19.31",
      },
      {
        "Id": 11,
        "EventTime": "2022-04-19T12:30:14.417",
      },
      {
        "Id": 16,
        "EventTime": "2022-04-19T12:38:16.267",
      },
      {
        "Id": 21,
        "EventTime": "2022-04-19T13:03:31.477",
      },
      {
        "Id": 21,
        "EventTime": "2022-04-19T13:03:32.657",
      },
      {
        "Id": 15,
        "EventTime": "2022-04-19T14:01:34.473",
      },
      {
        "Id": 27,
        "EventTime": "2022-04-22T10:29:09.873",
      }
    ]
  }]

我想要一个时间序列图表，显示与 PartId 对应的每个事件之间的时间。 所以事件栏应该占据整个轴直到下一个事件发生。

我尝试根据 PartId 对数据进行分组：

def GetEventPartIddf(df, iterations=2):

    df_strip = df["Events", "PartId"]].copy()

    event_dict = {"PartId": [], "EventTime": []}
    for i in range(iterations):
        el = df_strip.loc[i, "Events"]
        partId = df_strip.loc[i,"PartId"]
        for x in el:
            eventTime = pd.to_datetime(x["EventTime"])
            event_dict["EventTime"].append(eventTime)
            event_dict["PartId"].append(partId)
    df_events = pd.DataFrame.from_dict(event_dict)
    partIdGrp = df_events.groupby(["PartId"])
    return df_events, partIdGrp

但我不确定如何在 matplotlib 中打印此 plot。我尝试循环遍历 dataframe，但我无法使其工作。

def plotStackedBar(df, partIdGroup):
    fig, ax = plt.subplots(figsize=(16, 5))
    for partId, df_group in partIdGroup:
        event = []
        diffTime = []
        i = 0
        color = iter(cm.rainbow(np.linspace(0, 1, 10)))
        for row_index, row in df_group.iterrows():
            eventTime = row['EventTime']
            if (len(event) != 0):
                lastEventTime = event[i-1]
                diff = (eventTime - lastEventTime).total_seconds() / 60
                diffTime.append(diff)
                print(diff)
                if (diff > 3000):
                    break
            i += 1
            event.append(eventTime)
            c = next(color)
            ax.barh(partId, diffTime, color=c)
        print(event)
    plt.show()

Answer 1

更新：这是获取所需内容的完整代码。 我正在使用matplotlib.pyplot.eventplot 。

这个想法是以给定的频率生成一个虚拟的 dataframe（例如，在这种情况下每 1 分钟一行）。 然后，我通过附加每 1 分钟的时间数据来创建绘图数据，以防事件在此期间处于活动状态。

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

df = pd.DataFrame([{'PartId': dic.get('PartId'), **event} for dic in data for event in dic.get('Events',[])])
df['EventTime'] = pd.to_datetime(df['EventTime'])
df['NextEventTime'] = df.sort_values('EventTime').groupby('PartId').EventTime.shift(-1)
df['diff'] = df['NextEventTime'] - df['EventTime']
df['diff_hours'] = (df['diff']/ np.timedelta64(1, 'h')).astype(float)
df = df.dropna()

min_event_time = df.EventTime.min()
max_event_time = df.EventTime.max()
from datetime import datetime, timedelta
t = np.arange(min_event_time, max_event_time, timedelta(minutes=1)).astype(datetime)

plot_df = pd.DataFrame(t,columns=['Time'])

plot_data = tuple()
offsets = list()
colors = list()

for (partid, id), df_grouped in df.groupby(['PartId', 'Id']):
    data_for_id = []
    for j, df_row in df_grouped.iterrows():
        for i, row in plot_df.iterrows():
            if (row['Time'] >= df_row['EventTime']) & (row['Time'] <= df_row['NextEventTime']):
                data_for_id.append(row['Time'])
    if len(data_for_id) > 0:
        plot_data += (data_for_id,)
        offsets.append(partid)
        colors.append(f'C{id}')
        
plt.eventplot(positions=plot_data, orientation='horizontal', lineoffsets=offsets, linelengths=0.2, colors=colors, linestyles='solid')
plt.show()

此代码假定您的输入数据在字段data中。