Plotting bar chart of categorical values for each group-CodePudding

I have a dataframe that I want to group and then to plot (barplot) based on value_counts.

import pandas as pd

df = pd.DataFrame({'date': ['jan', 'jan', 'jan', 'jan', 'feb', 'feb', 'feb', 'feb'],
                  'value': ['low', 'low', 'high', 'medium', 'medium', 'high', 'high', 'medium']})
                  
print(df)

I want to have months on the X-axis (in this case only jan and feb) and bars of values for each month on Y-axis. Hight of the bars should depend of count of specific category. For example, for 'jan' low=2, high=1, medium=1

I have tried:

sns.countplot(x = 'date', hue = 'value', data = df.melt())

But this is my error:

ValueError: Could not interpret input 'date'

This can be done with matplotlib or seaborn

CodePudding user response：

You can use sns.countplot to count items from the original dataframe. You can use hue= to separate out the value column. order= can fix an order on the x-values. Similarly, hue_order= can set an order for the hue categories. (Default, the order of appearance in the dataframe is used.) palette= can among others be a dictionary to assign a specific color to a specific category.

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

df = pd.DataFrame({'date': ['jan', 'jan', 'jan', 'jan', 'feb', 'feb', 'feb', 'feb'],
                   'value': ['low', 'low', 'high', 'medium', 'medium', 'high', 'high', 'medium']})

sns.set()
ax = sns.countplot(data=df, x='date', hue='value',
                   # order=['jan', 'feb', 'mar', 'apr', 'may', 'jun', 'jul', 'aug', 'sep', 'oct', 'nov', 'dec'],
                   order=['jan', 'feb'], hue_order=['low', 'medium', 'high'],
                   palette={'low': 'limegreen', 'medium': 'gold', 'high': 'crimson'})
ax.set_xlabel('')  # optionally remove 'date' label, as it is clear from the ticklabels
plt.tight_layout()
plt.show()