Plotly categorical scatter/line plots in python

Question

First time posting here. I have been recently been introduced to seaborn and plotly python libraries. They produce some great plots.

I have been struggling recreating a seaborn plot in plotly. I have a dataframe that looks like this example: Example data set

In seaborn if I use the following code:

sns.catplot(data=test, x='time', y='y', hue='group', kind='point')

I get a great graph in which the groups have automatically been averaged and plots two lines with error bars. I have two lines on my graph of group 'a' and 'b' showing the average value over each timepoint: Seaborn example graph

If I try to replicate this in plotly using:

px.line(test,x='time', y='y', color='group')

I get the individual lines for each group (three for 'a' and three for 'b') like this:plotly example

Is there way to plot the averages in plotly, rather than the individual points. I can not figure out how to get plotly to plot the averages of each group like seaborn does.

Any help would be greatly appreciated.

Hope this makes sense.

Thanks

jayveesea jayveesea · Accepted Answer · 2020-05-06T13:40:59

With plotly you need to use transformations, in this case several, one to group them and then one to create the averages (with aggregate).

For instance, with some data:

import numpy as np
import pandas as pd

# some data
N=24
ds = pd.DataFrame({'group': np.random.choice(['a','b'],N),
                   'time': np.random.randint(0,6,N),
                   'y': np.random.randint(1,15,N)})
ds.sort_values(by=['group', 'time'],inplace=True)

then you can use plotly:

import plotly.io as pio

data = [dict(
  type = 'scatter', x = ds.time, y = ds.y,
  mode = 'lines+markers',
  transforms = [
      dict(
        type = 'groupby',
        groups = ds.group),
      dict(
        type = 'aggregate',
        groups = ds.time,
        aggregations = [dict(target = 'y', func = 'avg')]),
    ]
)]

fig_dict = dict(data=data)
pio.show(fig_dict, validate=False)

Plotly categorical scatter/line plots in python

1 Answers