Weighted bins in a distribution hist plot

Question

I'm looking for a way to plot a distribution histogram, with the y-axis representing the total number of items for each bin (and not just the count).

Example on the charts below:

• On the left, there are 55 agencies who sold between 20-30 houses
• On the right, the agencies having sold between 20-30 houses represent 1100 houses sold

It's not as trivial as it looks because one can't simply multiply each bin's count by the bin's value (maybe in the 20-30 bin, there are 54 agencies who sold 21 are 1 who sold 29).

Questions:

• What is the name of such a chart (the one on the right)?
• Is there a way to plot it natively in matplotlib or seaborn?

Answer 1

You want to use the weights kwarg (see numpy docs) which is passed through ax.hist (see).

Something like

fig, ax = plt.subplots()
ax.hist(num_sold, bins, weights=num_sold)

Answer 2

Edit: @tacaswell is better use it. But the labels for mine will line up correctly without hassle and the bars will be separated.

Hopefully your data is in pandas. I will create some fake data and then give you a solution.

import pandas as pd

# create a dataframe of number of homes sold
df = pd.DataFrame(data={'sold':np.random.randint(0,100, 1000)})

# groupby the left side of interval [0, 10), [10, 20) etc..  and plot
df.groupby(df.sold // 10 * 10).sum().plot.bar()