时间戳

首页 > 解决方案 > 未与 X 轴刻度对齐的条形和散景图中的第一个条形截止

pandas - 未与 X 轴刻度对齐的条形和散景图中的第一个条形截止

问题描述

我浏览了根据我的问题标题自动填充的帖子，但找不到与我的古怪结果完全一致的任何内容。

我正在输入数据包捕获的 CSV 文件并收集协议，分别对每个协议的每个协议的长度求和，然后除以协议总数以获得每个协议的平均数据包大小。

使用我的显示命令，我能够确认协议列和字节列具有相同的大小（有 18 个协议和 18 个字节大小的平均值）。除了可能被视为一个问题的两件事之外，我还能够确认该图表是否正确绘制。

  1. “DNS”的第一根柱子开始得早，半截断
  2. 条形图未在 x 轴刻度上对齐

我试图在他们的文档中使用简单的散景水果示例来复制这个问题，但它绘制正常，但这只是创建了一个带有 x 和 y 的虚拟数组。

偏心图

这是代码，但请记住，未附加 CSV 文件，并且由于 IP 地址的原因，我不得不避免分享它，但如果有人看到了任何建议或建议，我们将不胜感激。

此外，这是temp.head()的结果。

# Here is the myriad of declarations that helps this program work

import pandas as pd
import math as m
import ipympl
import ipywidgets
import numpy as np

# Bokeh libraries and modules (I am aware not all of these are required, just
# occasionally trying a few fun features, though let me know if this could be
# the problem)

from bokeh.io import  show, reset_output, output_notebook, export_png
from bokeh.plotting import figure, output_file
from bokeh.models import Range1d, FactorRange, ColumnDataSource, LabelSet, HoverTool
from bokeh.models import ColorBar, LogColorMapper, LogTicker
from bokeh.layouts import gridplot, row, column
from bokeh.transform import factor_cmap, linear_cmap
from bokeh.models.annotations import Label
from bokeh.models.tools import HoverTool
from bokeh.palettes import Spectral6, Category20, viridis, turbo, linear_palette

# Set up plots to stay inside the notebook
# Remove this when you want to bring up a separate window display

output_notebook()

# Set up Bokeh visualtization toolset  

TOOLS = "pan,wheel_zoom,box_zoom,reset,save"

data_r = pd.read_csv(r'C:\Users\xxx\Desktop\xxx.csv')
data_s = data_r.groupby('Protocol').Length.sum()
data_p = data_r.groupby('Protocol').Source.count()
data_a_p = data_s / data_p
data_a_p_df = data_a_p.to_frame()

temp = data_a_p_df.reset_index()
temp.columns = ['Protocol', 'bytes']

# Setting up the ColumnDataSource
cds = ColumnDataSource(temp)

p = figure(x_range=cds.data['Protocol'],
           plot_height=300,
           plot_width=800, 
           title="Average Packet Size by Protocol",
           y_axis_label='Size in Bytes',
           tools=TOOLS)

p.vbar(range(len(cds.data['Protocol'])), 
       width=.8,
       top=cds.data['bytes'],
       line_color='black',
       fill_color=turbo(len(cds.data['Protocol'])),
       fill_alpha=.5)

# To help the labels fit nicely, rotate the x-axis labels 45 degrees
p.xaxis.major_label_orientation = 45

# Display the graph
show(p)
display (len(cds.data['Protocol'])) # Results in 18
display (len(cds.data['bytes'])) # Results in 18

编辑：为了制作一个显示错误的功能示例，这里是使用虚拟数据框的更新代码：

# Here is the myriad of declarations that helps this program work
# Note I added prettytable to make the dummy csv file

import pandas as pd
import math as m
import ipympl
import ipywidgets
import numpy as np

# Bokeh libraries and modules (I am aware not all of these are required, just
# occasionally trying a few fun features, though let me know if this could be
# the problem)

from bokeh.io import  show, reset_output, output_notebook, export_png
from bokeh.plotting import figure, output_file
from bokeh.models import Range1d, FactorRange, ColumnDataSource, LabelSet, HoverTool
from bokeh.models import ColorBar, LogColorMapper, LogTicker
from bokeh.layouts import gridplot, row, column
from bokeh.transform import factor_cmap, linear_cmap
from bokeh.models.annotations import Label
from bokeh.models.tools import HoverTool
from bokeh.palettes import Spectral6, Category20, viridis, turbo, linear_palette

# Set up plots to stay inside the notebook
# Remove this when you want to bring up a separate window display

output_notebook()

# Set up Bokeh visualtization toolset  

TOOLS = "pan,wheel_zoom,box_zoom,reset,save"

# For this example I created a dummy dataframe for the data

data_r = pd.DataFrame({
   'Protocol': ['DNS','DNS', 'TCP', 'ICMPv6', 'TCP', 'TCP',
                 'HTTP', 'HTTP', 'TCP', 'TCP', 'TCP', 'TCP',
                 'TCP', 'ICMPv6', 'TCP', 'TCP', 'TCP', 
                 'ICMPv6', 'ICMPv6', 'AJP13', 'AJP13'],
   'Length': [96, 154, 66, 110, 171, 171, 208, 209, 56, 56,
            56, 56, 66, 110, 54, 55, 56, 110, 110, 171, 171]
})

data_s = data_r.groupby('Protocol').Length.sum()
data_p = data_r.groupby('Protocol').Protocol.count()
data_a_p = data_s / data_p
data_a_p_df = data_a_p.to_frame()

temp = data_a_p_df.reset_index()
temp.columns = ['Protocol', 'bytes']

# Setting up the ColumnDataSource
cds = ColumnDataSource(temp)

p = figure(x_range=cds.data['Protocol'],
           plot_height=300,
           plot_width=800, 
           title="Average Packet Size by Protocol",
           y_axis_label='Size in Bytes',
           tools=TOOLS)

p.vbar(range(len(cds.data['Protocol'])), 
       width=.8,
       top=cds.data['bytes'],
       line_color='black',
       fill_color=turbo(len(cds.data['Protocol'])),
       fill_alpha=.5)

# To help the labels fit nicely, rotate the x-axis labels 45 degrees
p.xaxis.major_label_orientation = 45

# Display the graph
show(p)

标签： pandasjupyter-notebookbar-chartbokeh

解决方案

首先，当您提供依赖于某些数据的代码时，请务必同时提供该数据。不是作为图片temp.head()，而是作为可以复制的东西。理想情况下，只需在代码本身中包含一些玩具数据。

至于你的问题 - 只是不要使用range(len(...))in p.vbar。只需提供cds.data['Protocol']作为第一个参数。

推荐阅读