Home How to get a grouped bar plot in Python in which one of the bar is stacked and the others are not?

Questions

How to get a grouped bar plot in Python in which one of the bar is stacked and the others are not?

July 9, 2022

I have a pandas dataframe df which looks as follows:

World   Coal    Oil and natural gas Nuclear Renewables  Storage Electricity networks
2015    79.6    79.8    28.4    309.9   1.6 316.8
2016    70.9    83.2    33.7    318.3   2.5 327.9
2017    65.8    77.2    36.9    325.9   3.1 318.6
2018    64.7    64.5    34.0    359.0   4.7 310.3
2019    62.2    67.4    34.8    393.2   4.6 291.8
2020    56.4    55.3    39.5    418.4   6.2 291.6
2021    51.6    67.0    43.7    445.8   9.4 308.1
2022    48.2    70.8    49.3    471.7   18.3    318.2

df.to_dict() is as shown below:

{'Coal': {2015: 79.6,
  2016: 70.9,
  2017: 65.8,
  2018: 64.7,
  2019: 62.2,
  2020: 56.4,
  2021: 51.6,
  2022: 48.2},
 'Oil and natural gas': {2015: 79.8,
  2016: 83.2,
  2017: 77.2,
  2018: 64.5,
  2019: 67.4,
  2020: 55.3,
  2021: 67.0,
  2022: 70.8},
 'Nuclear': {2015: 28.4,
  2016: 33.7,
  2017: 36.9,
  2018: 34.0,
  2019: 34.8,
  2020: 39.5,
  2021: 43.7,
  2022: 49.3},
 'Renewables': {2015: 309.9,
  2016: 318.3,
  2017: 325.9,
  2018: 359.0,
  2019: 393.2,
  2020: 418.4,
  2021: 445.8,
  2022: 471.7},
 'Storage': {2015: 1.6,
  2016: 2.5,
  2017: 3.1,
  2018: 4.7,
  2019: 4.6,
  2020: 6.2,
  2021: 9.4,
  2022: 18.3},
 'Electricity networks': {2015: 316.8,
  2016: 327.9,
  2017: 318.6,
  2018: 310.3,
  2019: 291.8,
  2020: 291.6,
  2021: 308.1,
  2022: 318.2}}

I am able to create a stacked bar plot using this entire data using

colors = ["black","gray","red","green","blue","yellow"]
df.plot(kind = "bar", stacked = True, color = colors)

The stacked bar plot is as shown below:

However, I’d like to have three separate bars for Generation, Storage and Electricity Networks. Generation here refers to data for Coal, Oil and natural gas, Nuclear and Renewables. How am I able to do it using matplotlib in Python?

>Solution :

You can plot first the stacked bars right aligned and then, on the same axes, the remaining bars left aligned.
You can then fix the legend to your liking.

ax = df.iloc[:,:4].plot.bar(stacked=True, align='edge', width=-.2, color=colors[:4])
df.iloc[:,4:].plot.bar(ax=ax, align='edge', width=.4, color=colors[4:])

handles, labels = ax.get_legend_handles_labels()
labels = [f'Generation ({l})' if i < 4 else l for i,l in enumerate(labels)]
ax.legend(handles, labels, bbox_to_anchor=(1, 1))