TypeError: %d format: a number is required, not numpy.float64

Question

Trying to plot, I got the following error from matplotlib:

TypeError: %d format: a number is required, not numpy.float64

This is the complete traceback (I've modified path names):

Traceback (most recent call last):
  File ".../plotmod.py", line 154, in _plot
    fig.autofmt_xdate()
  File ".../local/lib/python2.7/site-packages/matplotlib/figure.py", line 426, in autofmt_xdate
    for label in ax.get_xticklabels():
  File ".../local/lib/python2.7/site-packages/matplotlib/axes.py", line 2620, in get_xticklabels
    self.xaxis.get_ticklabels(minor=minor))
  File ".../local/lib/python2.7/site-packages/matplotlib/axis.py", line 1118, in get_ticklabels
    return self.get_majorticklabels()
  File ".../local/lib/python2.7/site-packages/matplotlib/axis.py", line 1102, in get_majorticklabels
    ticks = self.get_major_ticks()
  File ".../local/lib/python2.7/site-packages/matplotlib/axis.py", line 1201, in get_major_ticks
    numticks = len(self.get_major_locator()())
  File ".../local/lib/python2.7/site-packages/matplotlib/dates.py", line 595, in __call__
    'RRuleLocator estimated to generate %d ticks from %s to %s: exceeds Locator.MAXTICKS * 2 (%d) ' % (estimate, dmin, dmax, self.MAXTICKS * 2))
TypeError: %d format: a number is required, not numpy.float64

Where can this error come from?

Some basic research I've made results:

the error is not the real error, but instead one which is caused while trying to format the RuntimeError message that matplotlib.dates raises
the formatting error was due to python's %d, which, it seems, cannot handle numpy.float64 instances
the instance which has contained that data type is either estimate, which is some inner calculation result of matplotlib, or MAXTICKS, which is probably a constant, hence I tend to believe it's the first option
the calculation of estimate involves date2num which should return legitimate values, and _get_unit() and _get_interval(), which go deep enough into the module, and this is where my research stops.

I can easily reproduce the error in my entire software framework, but I can't isolate it for easy reproduction code. I think it tends to happen when the entire axis that should be plotted is very short (say, up to a few minutes long).

Any thoughts?

Looks like a matplotlib bug. You can ask on the matplotlib-users mailing list (lists.sourceforge.net/lists/listinfo/matplotlib-users), or create an issue on the matplotlib github site (github.com/matplotlib/matplotlib/issues). — Warren Weckesser
– Warren Weckesser, Commented Apr 8, 2014 at 12:14

dr jimbob · Accepted Answer · 2014-04-08 17:52:13Z

It seems you have a NaN or infinity that you are trying to format as an integer which raises the error (there's no such thing as a NaN or Inf for the int datatype).

In [1]: import numpy as np

In [2]: '%d' % np.float64(42)
Out[2]: '42'

In [3]: '%d' % np.float64('nan')
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)

<ipython console> in <module>()

TypeError: %d format: a number is required, not numpy.float64

In [4]: '%d' % np.float64('inf')
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)

ipython console> in <module>()

TypeError: %d format: a number is required, not numpy.float64

You could go into the matplotlib file (or use a python debugger) that generates the error and change the print line to have a %f which will work with all numpy floats. ('%f' % np.float64('nan') returns 'nan').

user297171 · Accepted Answer · 2014-04-08 12:01:49Z

0

Convert numpy float to Python float manually.

np.asscalar(np.float64(42))

answered Apr 8, 2014 at 12:01

user297171

2 Comments

Bach Over a year ago

1. Since the numbers appear inside the matplotlib module, I can't do that without editing it. I wish to avoid that, as this is an external module.

Bach Over a year ago

2. This will solve the formatting error, but as I wrote, this is not the real error - this will only format the RuntimeError properly.

Collectives™ on Stack Overflow

TypeError: %d format: a number is required, not numpy.float64

2 Answers 2

Comments

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related