How to create scatter plot + bubble chart?

tajinder.dhillon

Hello,

Can you please help me amend my code based on the following questions:

1. How can I create a scatter plot showing Gross margin on x-axis and EBITDA margin on y-axis?

2. How can I create a BUBBLE chart showing the same data in #1 above, where the bubble is defined by market cap in USD?

Thank you

-----------------------------------------------

ric = 'AAPL.O'

df, err = ek.get_data([ric,"Peers("+ric+")"],

['TR.CompanyMarketCap',

'TR.RevenueMean',

'TR.GrossIncomeMean',

'TR.EBITDAMean'],

parameters = {

'Scale': '6',

'Curn': 'USD'

})

df

df.columns = ['RIC', 'Market Cap', 'Revenue', 'Gross Income', 'EBITDA']

df['Gross Profit Margin'] = df['Gross Income']/df['Revenue']

df['EBITDA Margin'] = df['EBITDA']/df['Revenue']

display(df)

zoya faberov

Hello @tajinder.dhillon,

You can use scatter-plotting that is part of Pandas DataFrame:

df.plot.scatter(x = 'Gross Profit Margin', y = 'EBITDA Margin')

should so a scatter plot, you can find reference here:

Pandas documentation: pandas.DataFrame.plot.scatter

---

To create bubble charts, you may wish to use Plotly (that is also included in CodeBook), you will need to convert the column types to numerics, in order for them to be charted, for example:

import plotly.express as px

df['Market Cap Num'] = df['Market Cap'].astype(float)
df['EBITDA Margin Num'] = df['EBITDA Margin'].astype(float)
df['Gross Profit Margin Num'] = df['Gross Profit Margin'].astype(float)
df

and

fig = px.scatter(df, 
    x = 'Gross Profit Margin Num', 
    y = 'EBITDA Margin Num',
size="Market Cap Num", #Third Parameter
    log_x=True, size_max=60)
fig.show()

yields:

tajinder.dhillon

many thanks @zoya faberov Zoya that worked. For the bubble chart, how would I be able to do the following (if you could amend the code so I can paste it that would be great):

1. Change the colour of the bubbles using R,G,B (i.e. 0, 30, 255).
2. Change the colour of a single bubble (i.e focus on single security to differentiate from peers)
3. Display the RIC for each bubble.
4. Display the RIC, x-value, y-value (helpful if the universe is small).
5. Is there a way to make the actual chart larger so the bubbles have more room/make the chart easier to analyze quadrants?
6. Are we able to use a heatmap colour scheme for the bubbles (i.e. red (values in bottom left quadrant) to green (values in top right quadrant)?)

Kind regards

zoya faberov

Hello @tajinder.dhillon,

You can find these features and much more on plotly bubble charts on many external python learning resources, for example:

https://pythonwife.com/bubble-chart-with-plotly/

https://www.geeksforgeeks.org/bubble-chart-using-plotly-in-python/

https://plotly.com/python/bubble-charts/

Moving up from the super-quick bubble charts allowing to grasp the data and plotly.express module, to more customizable charts allowing for flexibility of expression, often involves the use of plotly.graph_objects module, as you will find in many of the included examples.

Hope that this information is of help to you?

tajinder.dhillon

Thank you @zoya faberov for the links.

Perhaps if you can share code for the below that will be OK for now as these are more important for my use case:

1. Physically display the RIC for each bubble.
2. Physically display the RIC, x-value, y-value (helpful if the universe is small).
3. Change the colour of a single bubble (i.e focus on single security to differentiate from peers)

From one of the links you provided, I was able to incorporate hover_data=['RIC'], but this shows the RIC, however, only if I hover a bubble as opposed to seeing it physically on the chart.

Kind regards

zoya faberov

Hello @tajinder.dhillon ,

The simplest implementation, in my view, that I think may work is something like:

fig = px.scatter(df, 
    x = 'Gross Profit Margin Num', 
    y = 'EBITDA Margin Num',
    size="Market Cap Num", 
    text= df['RIC'],
    color = "Market Cap Num",
    log_x=True, 
    size_max=100)
fig.show()

That yields for me a chart like:

Not the prettiest and too busy for my test, but with the largest and the smallest of the values clearly visible because of the color scale.

zoya faberov

Hello @tajinder.dhillon ,

I have also worked on changing the colour of a single bubble (focus on single security to differentiate from peers), here is my take:

1. Prepare numeric and adjusted values for columns

df['Market Cap Num'] = (df['Market Cap']).astype(float)
df['Market Cap Num Adj'] = (df['Market Cap']/10000).astype(int)
df['EBITDA Margin Num'] = df['EBITDA Margin'].astype(float)
df['Gross Profit Margin Num'] = df['Gross Profit Margin'].astype(float)
df

2. Create array of the same blue RGB value objects, and put in a single marker of RGB red at specific position:

import numpy as np

colorArray= np.full(
  shape=len(df['EBITDA Margin Num']),
  fill_value='rgb(93, 164, 214)',
   dtype=object
)

colorArray[20] = 'rgb(255, 65, 54)'
#colorArray

3. Chart with Plotly Graph Objects:

import plotly.graph_objects as go


fig = go.Figure(data=[go.Scatter(
    x=df['Gross Profit Margin Num'], 
    y=df['EBITDA Margin Num'],
    text=df['RIC'],
    mode='markers+text',
    marker=dict(
        color=colorArray,
        size=df['Market Cap Num Adj'],
    ),
)])

fig.show()

The results on my side look like this:

I hope this is of help

tajinder.dhillon

thank you @zoya faberov , is there a way to implement a "max_size=100" in the below instead of having to divide market cap by an arbitrary number?

fig = go.Figure(data=[go.Scatter(

x=df['Gross Profit Margin %'],

y=df['EBITDA Margin %'],

text=df['RIC'],

mode='markers+text',

marker=dict(

color=colorArray,

size=df['Market Cap Num Adj'],

zoya faberov

Hello @tajinder.dhillon ,

Yes you can- slightly more code:

import plotly.graph_objects as go

df['Market Cap Num'] = (df['Market Cap']).astype(int)

fig = go.Figure(data=[go.Scatter(
    x=df['Gross Profit Margin Num'], 
    y=df['EBITDA Margin Num'],
    text=df['RIC'],
    mode='markers+text',
    marker=dict(
        color=colorArray,
        size=df['Market Cap Num'],
        sizemode='area',
        sizeref=2.*max(df['Market Cap Num'])/(40.**2),
        sizemin=4
    ),
)])

fig.show()

Resulting in:

... however, in my opinion, visual representations of content are always like this- a simple one can be put together quite quickly, but really good ones have to take care of every detail, and end up taking time to get them right