pandas plot multiple columns bar chart - grouped and stacked chart

In this article, we will see how to create a grouped bar chart and stacked chart using multiple columns of a pandas dataframe

Here are the steps that we will follow in this article to build this multiple column bar chart using seaborn and pandas plot function

• Create a test dataframe
• Build a grouped bar chart using pandas plot function
• Create a pivot table to create a stacked bar chart
• Build a multiple column bar chart using seaborn

Create a dataframe

We will first create a test dataframe with monetary details for an year. It has got four columns - month, sales, tax and profit.

df=pd.DataFrame(
                {'month': 
                         ['jan', 'feb', 
                          'mar', 'apr', 
                          'may', 'jun', 
                          'jul', 'aug', 
                          'sep', 'oct', 
                          'nov', 'dec'],
              'sales': [45, 13, 28, 32, 
                        40, 39, 26, 35, 
                        22, 18, 42, 30],
              'tax': [5, 2, 4, 6, 8, 7, 
                           3, 5, 3, 2, 10, 6],
              'profit': [40, 11, 24, 26, 32, 32, 
                         23, 30, 32, 20, 8, 36]})

df

This is how our test dataframe looks like:

	Month	Sales	Tax	Profit
0	jan	45	5	40
1	feb	13	2	11
2	mar	28	4	24
3	apr	32	6	26
4	may	40	8	32
5	jun	39	7	32
6	jul	26	3	23
7	aug	35	5	30
8	sep	22	3	32
9	oct	18	2	20
10	nov	42	10	8
11	dec	30	6	36

Create a grouped bar chat with multiple columns

Pandas plot:

We will use pandas plot function and pass month column as x parameter and all other columns as list to y parameter

(df.plot(
        x='month', 
        y=['sales','profit', 'tax-paid'], 
        kind='bar', 
        figsize=(5,5))
        .legend( bbox_to_anchor =(1 ,1)
       )
)

Horizontal bar plot:

Update the kind parameter to barh to create a horizontal bar chart

(df.plot(
          x='month', 
          y=['sales','profit', 'tax-paid'], 
          kind='barh', 
          figsize=(5,5))
        .legend( bbox_to_anchor =(1 ,1)
       )
)

Pivot table plot:

We could also create the grouped bar chart with multiple columns by first creating a pivot table from the dataframe and then plot it

(
  pd.pivot_table(
               df, 
               index=['month'], 
               sort=False)
              .plot(kind='bar', 
                    figsize=(5,5))
              .legend( bbox_to_anchor =(1 ,1)
               )
)

Create a stacked bar chat

Just in case, you would like to plot the stacked bar chart of all those columns instead of a grouped bar chart, we could just add a stacked parameter in the pandas plot function to built it

(
  pd.pivot_table(
               df, 
               index=['month'], 
               sort=False)
              .plot(kind='bar', 
                    figsize=(5,5),
                    stacked = True)
              .legend( bbox_to_anchor =(1 ,1)
               )
)

#OR

(
  df.plot(
          x='month', 
          y=['sales','profit', 'tax-paid'], 
          kind='bar', 
          figsize=(5,5), 
          stacked=True)
        .legend( bbox_to_anchor =(1 ,1)
         )
)

Create a grouped bar chat using seaborn

Seaborn provides some easy to plot grouped bar charts functions, we need to first reshape the dataframe and melt it so that we have a dataframe in long format as shown here

df1=pd.melt(
            df, 
            id_vars="month", 
            var_name="revenue", 
            value_name="amount"
      )
df1

	Month	Accounts_category	Amount
0	jan	sales	45
1	jan	tax-paid	5
2	jan	profit	40
3	feb	sales	13
4	feb	tax-paid	2
5	feb	profit	11
6	mar	sales	28
7	mar	tax-paid	4
8	mar	profit	24
…	…	…	…

To plot a grouped bar chart, we could use either seaborn barplot or catplot

fig, ax = plt.subplots(figsize=(8, 8), dpi=100)

sns.barplot(
              x='month', 
              y='amount', 
              hue='revenue', 
              data=df1,  
              ax=ax
            )

# OR

sns.catplot(
            x='month', 
            y='amount', 
            hue='revenue', 
            data=df1, 
            kind='bar'
          )