How to make multiple plots with seaborn from a wide dataframe

Question

I'm currently learning about data visualization using seaborn, and I came across a problem that I couldn't find a solution to.

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

So I have this data

index	col1	col2	col3	col4	col5	col6	col7	col8
1990	0	4	7	3	7	0	6	6
1991	1	7	5	0	8	1	8	4
1992	0	5	0	1	9	1	7	2
1993	2	7	0	0	6	1	2	7
1994	4	1	5	5	8	1	6	3
1995	7	0	6	4	8	0	5	7
1996	5	1	1	4	6	1	7	4
1997	0	4	7	5	5	1	8	5
1998	1	3	7	0	7	0	7	1
1999	5	7	1	1	6	0	8	5
2000	3	8	5	0	3	0	6	3
2001	6	0	4	1	7	1	2	7

I want to make barplots/histplots with col1, col2 .. col8 as one column and 1990 values as one column so like 1990;

col?	val
col1	0
col2	4
col3	7
col4	3
col5	7
col6	0
col7	6
col8	6

and plot them for each year from 1990 to 2001.

g = sns.FacetGrid(df, col=df.index.value_counts())
g.map(sns.histplot, df.columns)

This is the code that I've written I looked at facetgrid but could get it working for my case, any feedback is appreciated.

Answer 1

Imports and Test DataFrame

• Tested with pandas 1.3.0, matplotlib 3.4.2, and seaborn 0.11.1

import pandas as pd
import seaborn as sns

# sample dataframe
data = {1990: {'col1': 0, 'col2': 4, 'col3': 7, 'col4': 3, 'col5': 7, 'col6': 0, 'col7': 6, 'col8': 6}, 1991: {'col1': 1, 'col2': 7, 'col3': 5, 'col4': 0, 'col5': 8, 'col6': 1, 'col7': 8, 'col8': 4}, 1992: {'col1': 0, 'col2': 5, 'col3': 0, 'col4': 1, 'col5': 9, 'col6': 1, 'col7': 7, 'col8': 2}, 1993: {'col1': 2, 'col2': 7, 'col3': 0, 'col4': 0, 'col5': 6, 'col6': 1, 'col7': 2, 'col8': 7}, 1994: {'col1': 4, 'col2': 1, 'col3': 5, 'col4': 5, 'col5': 8, 'col6': 1, 'col7': 6, 'col8': 3}, 1995: {'col1': 7, 'col2': 0, 'col3': 6, 'col4': 4, 'col5': 8, 'col6': 0, 'col7': 5, 'col8': 7}, 1996: {'col1': 5, 'col2': 1, 'col3': 1, 'col4': 4, 'col5': 6, 'col6': 1, 'col7': 7, 'col8': 4}, 1997: {'col1': 0, 'col2': 4, 'col3': 7, 'col4': 5, 'col5': 5, 'col6': 1, 'col7': 8, 'col8': 5}, 1998: {'col1': 1, 'col2': 3, 'col3': 7, 'col4': 0, 'col5': 7, 'col6': 0, 'col7': 7, 'col8': 1}, 1999: {'col1': 5, 'col2': 7, 'col3': 1, 'col4': 1, 'col5': 6, 'col6': 0, 'col7': 8, 'col8': 5}, 2000: {'col1': 3, 'col2': 8, 'col3': 5, 'col4': 0, 'col5': 3, 'col6': 0, 'col7': 6, 'col8': 3}, 2001: {'col1': 6, 'col2': 0, 'col3': 4, 'col4': 1, 'col5': 7, 'col6': 1, 'col7': 2, 'col8': 7}}
df = pd.DataFrame.from_dict(data, orient='index')

# display(df.head())
      col1  col2  col3  col4  col5  col6  col7  col8
1990     0     4     7     3     7     0     6     6
1991     1     7     5     0     8     1     8     4
1992     0     5     0     1     9     1     7     2
1993     2     7     0     0     6     1     2     7
1994     4     1     5     5     8     1     6     3

Plotting with seaborn.catplot

• Using seaborn 0.11.1, the correct way to create a barplot FacetGrid (per the documentation), is with sns.catplot and kind='bar'.
• It is required to convert the dataframe from a wide to long form, which is easily done by resetting the index, and then using pandas.DataFrame.melt()
• A catplot is a figure-level interface for drawing categorical plots onto a FacetGrid.
  • g.set_xticklabels(rotation=90) can be used to rotate the xticklabels.
  • See How to rotate xticklabels in a seaborn catplot or How to set rotation for seaborn FacetGrid and figure-level xtick labels

# convert the wide dataframe to a long format with melt
dfm = df.reset_index().melt(id_vars='index', var_name='variable', value_name='value')

# display(dfm.head())
   index variable  value
0   1990     col1      0
1   1991     col1      1
2   1992     col1      0
3   1993     col1      2
4   1994     col1      4

# plot with catplot and kind='bar'
g = sns.catplot(data=dfm, kind='bar', col='index', col_wrap=4, x='variable', y='value', height=3)

# change the ticklabel rotation if needed
g.set_xticklabels(rotation=90)

# change ylim if needed
g.set(ylim=(0, 30))

Plotting with pandas.DataFrame.plot

• While you have asked about seaborn, given the dataframe in the OP with all the years in the index, the easiest way to plot the data is transpose the dataframe with .T, and then use pandas.DataFrame.plot

# display(df.T.head())
      1990  1991  1992  1993  1994  1995  1996  1997  1998  1999  2000  2001
col1     0     1     0     2     4     7     5     0     1     5     3     6
col2     4     7     5     7     1     0     1     4     3     7     8     0
col3     7     5     0     0     5     6     1     7     7     1     5     4
col4     3     0     1     0     5     4     4     5     0     1     0     1
col5     7     8     9     6     8     8     6     5     7     6     3     7

# transpose and plot
axes = df.T.plot(kind='bar', subplots=True, layout=[3, 4], figsize=(15, 7), legend=False, rot=0)

# to change ylim of the subplots, if needed
for ax in axes.flatten():
    ax.set_ylim(0, 30)

Answer 2

melt your dataframe first.

df = pd.DataFrame({'index': {0: 1990, 1: 1991, 2: 1992, 3: 1993, 4: 1994, 5: 1995, 6: 1996, 7: 1997, 8: 1998, 9: 1999, 10: 2000, 11: 2001}, 'col1': {0: 0, 1: 1, 2: 0, 3: 2, 4: 4, 5: 7, 6: 5, 7: 0, 8: 1, 9: 5, 10: 3, 11: 6}, 'col2': {0: 4, 1: 7, 2: 5, 3: 7, 4: 1, 5: 0, 6: 1, 7: 4, 8: 3, 9: 7, 10: 8, 11: 0}, 'col3': {0: 7, 1: 5, 2: 0, 3: 0, 4: 5, 5: 6, 6: 1, 7: 7, 8: 7, 9: 1, 10: 5, 11: 4}, 'col4': {0: 3, 1: 0, 2: 1, 3: 0, 4: 5, 5: 4, 6: 4, 7: 5, 8: 0, 9: 1, 10: 0, 11: 1}, 'col5': {0: 7, 1: 8, 2: 9, 3: 6, 4: 8, 5: 8, 6: 6, 7: 5, 8: 7, 9: 6, 10: 3, 11: 7}, 'col6': {0: 0, 1: 1, 2: 1, 3: 1, 4: 1, 5: 0, 6: 1, 7: 1, 8: 0, 9: 0, 10: 0, 11: 1}, 'col7': {0: 6, 1: 8, 2: 7, 3: 2, 4: 6, 5: 5, 6: 7, 7: 8, 8: 7, 9: 8, 10: 6, 11: 2}, 'col8': {0: 6, 1: 4, 2: 2, 3: 7, 4: 3, 5: 7, 6: 4, 7: 5, 8: 1, 9: 5, 10: 3, 11: 7}})
df2 = df.melt(id_vars='index')

g = sns.FacetGrid(data=df2, col='index', col_wrap=4)
g.map(sns.barplot, 'variable', 'value', order=df2['variable'].unique())
plt.show()