pandas - 分组加权条形图

Question

Consider the following DataFrame consisting of 10 rows.

d = {
    'grp_id':[1,2,1,1,1,3,1,1,4,1],
    'weight':[1,2,1,1,1,3,1,1,4,4],
    'value': [1,2,1,3,2,1,4,1,1,3]
}
df = pd.DataFrame(d)

A weighted histogram can be achieved with

df['value'].hist(histtype='bar', weights=df['weight'])

An unweighted bar chart grouped by grp_id with

df['value'].hist(by=df['grp_id'], histtype='bar')

I'd like to combine the two and plot a weighted bar chart grouped by grp_id .
I've tried the following 2 methods without success since for both I get a ValueError .

df['value'].hist(by=df['grp_id'], weights=df['weight'], histtype='bar')
df['value'].hist(by=df['grp_id'], weights='weight', histtype='bar')

ValueError: weights should have the same shape as x

The temporary solution I'm using is the following.

fig, axes = plt.subplots(2, 2)
for ax,(idx, grp) in zip(axes.flatten(), df.groupby('grp_id')):
    grp['value'].hist(weights=grp['weight'], histtype='bar', ax=ax)

However, I would like to ask if there is a direct way to do it with pandas.

Answer 1

I would firstly create a new data frame storing weighted values:

df['weighted_values'] = df.weight*df.value
df = df.groupby('grp_id')['weighted_values'].sum().to_frame().reset_index()

You can use seaborn to plot the final barchart aesthetically:

import seaborn as sns
sns.barplot(x = 'grp_id', y = 'weighted_values', data=df)