Search code examples
pythonmatplotlibhistogram

How to Stop Matplotlib Auto-Formatting of ytick?

I am making histograms of several different sets of data, but in this data set which is a list of complexities, my yticks are getting auto-formatted. How do I get my yticks to be uniform?

def make_histogram(listOfAllComplexities): 
    plt.hist(listOfAllComplexities,ec='black',color='orange',log=True, 
    density=True,bins=5)
    plt.xlabel(r'$\tilde{K}(x)$')
    plt.ylabel('Frequency',labelpad=20)
    plt.xticks(size = 8)
    plt.yticks(size = 8)
    plt.title('Hist_{0}chars_{1}_{2}'.format(windowSize,func.__name__,data_short_name))
    plt.show()

make_histogram([7.0, 8.1, 7.0, 7.0, 9.3, 7.0, 8.1, 9.3, 7.0, 7.0, 7.0, 5.8, 7.0, 8.1, 9.3, 7.0, 8.1, 7.0, 5.8, 9.3, 5.8, 7.0, 7.0, 8.1, 8.1, 7.0, 8.1, 2.3, 7.0, 5.8, 8.1, 2.3])

You can see the resulting figure below. The problematic ytick is the second ytick from the bottom.

Histogram yticks are not uniform in style:
enter image description here

Solution

  • As mentioned in my comment, there are two possible problems. One is that you are puzzled why only some of the labels are changed in size. Matplotlib differentiates between major and minor ticks, and your approach only modifies the major y-ticks. This is easily resolved by accessing the axis object with ax.tick_params():

    import matplotlib.pyplot as plt
 
def make_histogram(listOfAllComplexities): 
    plt.hist(listOfAllComplexities,ec='black',color='orange',log=True, density=True,bins=5)
    plt.xlabel(r'$\tilde{K}(x)$')
    plt.ylabel('Frequency',labelpad=20)
    ax = plt.gca()
    ax.tick_params(axis="both", which="both", labelsize=8)
    plt.title('Hist_{0}chars_{1}_{2}'.format(" A ", " B ", " C "))
    plt.show()

make_histogram([7.0, 8.1, 7.0, 7.0, 9.3, 7.0, 8.1, 9.3, 7.0, 7.0, 7.0, 5.8, 7.0, 8.1, 9.3, 7.0, 8.1, 7.0, 5.8, 9.3, 5.8, 7.0, 7.0, 8.1, 8.1, 7.0, 8.1, 2.3, 7.0, 5.8, 8.1, 2.3])

    Sample output: enter image description here

    As for the formatting of the 10^-1, I assume this is a conscious decision by the matplotlib team, so one can clearly see the decades of the major ticks. However, we can build our own FuncFormatter to imitate the style used for the minor ticks:

    import matplotlib.pyplot as plt
import matplotlib.ticker as tkr  
import math


def numfmt(x, pos):
    sign_string = ""
    if x<0:
        sign_string = "-" 
        x = math.fabs(x)
    if x == 0:
        return r'$\mathdefault{0}$'
    base = 10
    exponent = math.floor(math.log10(x))
    coeff = round(x / (base ** exponent))
    return r'$\mathdefault{%s%g\times%s^{%d}}$' % (sign_string, coeff, base, exponent)

myfmt = tkr.FuncFormatter(numfmt)
 
def make_histogram(listOfAllComplexities): 
    plt.hist(listOfAllComplexities,ec='black',color='orange',log=True, density=True,bins=5)
    plt.xlabel(r'$\tilde{K}(x)$')
    plt.ylabel('Frequency',labelpad=20)
    ax = plt.gca()
    ax.tick_params(axis="both", which="both", labelsize=8)
    ax.yaxis.set_major_formatter(myfmt)
    plt.title('Hist_{0}chars_{1}_{2}'.format(" A ", " B ", " C "))
    plt.show()

make_histogram([7.0, 8.1, 7.0, 7.0, 9.3, 7.0, 8.1, 9.3, 7.0, 7.0, 7.0, 5.8, 7.0, 8.1, 9.3, 7.0, 8.1, 7.0, 5.8, 9.3, 5.8, 7.0, 7.0, 8.1, 8.1, 7.0, 8.1, 2.3, 7.0, 5.8, 8.1, 2.3])

    Sample output: enter image description here

    The FuncFormatter function numfmt() is over the top because I only retrospectively noticed that we do not need it for the minor ticks (let matplotlib take care of them) and your histogram frequencies will always be positive. Oh well.