Sorting Box Plots by Median using Plotly Graph Objects

Question

I'm pretty much a beginner in plotly/pandas/data but I'm trying to make this graph and no matter what I search up, I can't find any attributes that are compatible with dictionaries. The data I'm using is the Time series download speed for 9 different software. I am trying to display the box plot descending by their median values.

Here is my code:

import pandas as pd
import plotly.graph_objs as go
from plotly.offline import plot
import numpy as np
olddf = pd.read_csv("justice.csv")
df = olddf.interpolate()



col = df.loc[:,'Bfy':'Sfy']
df["1"] = col.mean(axis=1)
col2 = df.loc[:,'Bakamai':'Sakamai']
df["2"] = col2.mean(axis=1)
col4 = df.loc[:,'Bazure':'Sazure']
df["4"] = col4.mean(axis=1)
col5 = df.loc[:,'Bcloudflare':'Scloudflare']
df["5"] = col5.mean(axis=1)
col6 = df.loc[:,'Bfastly':'Sfastly']
df["6"] = col6.mean(axis=1)
col7 = df.loc[:,'BAWS':'SAWS']
df["7"] = col7.mean(axis=1)
col8 = df.loc[:,'Bali':'Sali']
df["8"] = col8.mean(axis=1)
col9 = df.loc[:,'Bgoog':'Sgoog']
df["9"] = col9.mean(axis=1)

trace_one = go.Box(
    y=df['1'],
    name="Fy",
    line = dict(color='#8235EA'),
    opacity = 0.8)
trace_two = go.Box(
    y=df['2'],
    name="Akamai",
    line = dict(color='#EA8933'),
    opacity = 0.8)
trace_four = go.Box(
    y=df['4'],
    name="Azure",
    line = dict(color='#62F92C'),
    opacity = 0.8)
trace_five = go.Box(
    y=df['5'],
    name="Cloudflare",
    line = dict(color='#3548EA'),
    opacity = 0.8)
trace_six = go.Box(
    y=df['6'],
    name="Fastly",
    line = dict(color='#D735EA'),
    opacity = 0.8)
trace_seven = go.Box(
    y=df['7'],
    name="AWS Cloudfront",
    line = dict(color='#29E5B7'),
    opacity = 0.8)
trace_eight = go.Box(
    y=df['8'],
    name="Alibaba Cloud",
    line = dict(color='#3597EA'),
    opacity = 0.8)
trace_nine = go.Box(
    y=df['9'],
    name="Google Cloud",
    line = dict(color='#EA4833'),
    opacity = 0.8,
    )
data=[trace_one, trace_four, trace_seven, trace_eight, trace_nine, trace_five, trace_two]

layout = dict(
    
        title = "CHINA - Software vs Mb loaded per second")

fig = dict(data=data, layout=layout)

plot(fig)

csv layout example:

datetime,Bfy,Sfy,Gfy,Bakamai,Sakamai,Gakamai,Bazuaka,Sazuaka,Gazuaka,Bazure,Sazure,Gazure,Bcloudflare,Scloudflare,Gcloudflare,Bfastly,Sfastly,Gfastly,BAWS,SAWS,GAWS,Bali,Sali,Gali,Bgoog,Sgoog,Ggoog
23/07/21 10:02PM,,,215200,1489,1571,,1897,12400,173600,6551,,,1556,769,,,,749,6124,9347,2179,4160,,4473,4635,906,3426
23/07/21 10:12PM,22653,21520,,,1670,,17360,,,,10850,,,18261,1522,,3414,2010,5148,10447,2030,2667,4160,4119,5837,1592,3216
23/07/21 10:22PM,23911,,,1535,1615,815,3156,13354,177,6313,,,,825,586,873,,885,4280,6458,2114,4039,4119,6303,5629,1072,3283

Please provide the dataset, or at least a dummy example. – mozway Aug 05 '21 at 07:51 — mozway, Aug 05 '21 at 07:51
hi i added a bit, is it sufficient? – Bea Aug 05 '21 at 08:30 — Bea, Aug 05 '21 at 08:30

score 1 · Answer 1 · answered Aug 05 '21 at 13:31

taken a different approach to data preparation
1. pair columns, calculate means
2. create new dataframe from these paired column means
order columns of this data preparation based on their medians
create box plots in same order as ordered columns
found two providers that your code did not plot...

import plotly.graph_objects as go
import pandas as pd
import io

df = pd.read_csv(io.StringIO("""datetime,Bfy,Sfy,Gfy,Bakamai,Sakamai,Gakamai,Bazuaka,Sazuaka,Gazuaka,Bazure,Sazure,Gazure,Bcloudflare,Scloudflare,Gcloudflare,Bfastly,Sfastly,Gfastly,BAWS,SAWS,GAWS,Bali,Sali,Gali,Bgoog,Sgoog,Ggoog
23/07/21 10:02PM,,,215200,1489,1571,,1897,12400,173600,6551,,,1556,769,,,,749,6124,9347,2179,4160,,4473,4635,906,3426
23/07/21 10:12PM,22653,21520,,,1670,,17360,,,,10850,,,18261,1522,,3414,2010,5148,10447,2030,2667,4160,4119,5837,1592,3216
23/07/21 10:22PM,23911,,,1535,1615,815,3156,13354,177,6313,,,,825,586,873,,885,4280,6458,2114,4039,4119,6303,5629,1072,3283"""))

# different approach to getting means per provider to plot
df2 = pd.DataFrame({c[1:]:df.loc[:,[c, "S"+c[1:]]].mean(axis=1).values for c in df.columns if c[0]=="B"})

# re-order columns on ascending median
df2 = df2.reindex(df2.median().sort_values().index, axis=1)

meta = {'fy': {'color': '#8235EA', 'name': 'Fy'},
 'azure': {'color': '#62F92C', 'name': 'Azure'},
 'AWS': {'color': '#29E5B7', 'name': 'AWS Cloudfront'},
 'ali': {'color': '#3597EA', 'name': 'Alibaba Cloud'},
 'goog': {'color': '#EA4833', 'name': 'Google Cloud'},
 'cloudflare': {'color': '#3548EA', 'name': 'Cloudflare'},
 'akamai': {'color': '#EA8933', 'name': 'Akamai'},
        # next two were missing
 'fastly': {'color': 'pink', 'name': 'Fastly'},
 'azuaka': {'color': 'purple', 'name': 'azuaka'},
       }

go.Figure([go.Box(y=df2[c], name=meta[c]["name"], line={"color":meta[c]["color"]}) for c in df2.columns])

Sorting Box Plots by Median using Plotly Graph Objects

1 Answers1