matplotlib hist(): weights should have the same shape as x while shape is the same

Question

I'm trying to plot a histogram of a column in a pandas series ('df_plot'). Since I want the y-axis to be a percentage (and not the count), I use the weights option achieve this. As you find in the stacktrace below, the weights array and data series are of the same shape. How come that I still get the error telling me w and x are not the same shape?

Code:

w = 100*(np.zeros_like(df_plot[var]) + 1. / len(df_plot[var]))
print w.shape
print df_plot[var].shape
df_plot[var].hist(bins=100,  cumulative=True, weights=w)

Stacktrace:

(9066,)
(9066,)


Traceback (most recent call last):

  File "<ipython-input-59-5612307b159e>", line 4, in <module>
    df_plot[var].hist(bins=100,  cumulative=True, weights=w)

  File "C:\Anaconda\lib\site-packages\pandas\tools\plotting.py", line 2819, in hist_series
    ax.hist(values, bins=bins, **kwds)

  File "C:\Anaconda\lib\site-packages\matplotlib\axes\_axes.py", line 5649, in hist
    'weights should have the same shape as x')

ValueError: weights should have the same shape as x

Answer 1

you have nulls in your data set.

s = df_plot[var].dropna()
w = 100*(np.zeros_like(s) + 1. / len(s))
s.hist(bins=100,  cumulative=True, weights=w)