How can I adapt the autolabel function in matplotlib so that it displays negative values correctly?

Question

I have been playing around with Python for the last couple of days and found a lot of good resources about labelling, but I am failing to make it properly display negative values. Because the autolabel() function takes the height of the bar, which seems to always be a positive value, the labels are displayed way up in the graph and are of course not displayed as negative values. Can I somehow get the values that make up these bars or how do I get these labels down where they belong and show them as negative?

import pandas as pd
import matplotlib.pyplot as plt
from builtins import list
import matplotlib
matplotlib.style.use('ggplot')
import numpy as np

n_groups = 2

# create plot
fig, ax = plt.subplots()
fig.canvas.set_window_title('Mindestlohn Bundesweit')
index = np.arange(n_groups)
bar_width = 0.20
opacity = 0.8

list_reallohn_week_vollzeit = [-8.159698443426123, 11.395025597733763]
list_reallohn_week_teilzeit = [-1.048913873322391, 28.99318154295449]
list_reallohn_week_mini = [-7.552596893170488, 7.959096278017519]


rects1 = plt.bar(index + 0.00, list_reallohn_week_vollzeit, bar_width,
             alpha=opacity,
             color='b',
             label='Vollzeit')
rects2 = plt.bar(index + bar_width, list_reallohn_week_teilzeit, bar_width,
             alpha=opacity,
             color='g',
             label='Teilzeit')
rects3 = plt.bar(index + bar_width * 2,list_reallohn_week_mini, bar_width,
             alpha = opacity,
             color='c',
             label='Mini Job')

label_week_lists = ('2014 vor MdL', '2015 Nicht MdL berechtigt', '2015 mit MdL')

plt.ylabel('EUR')
plt.title('Reallöhne pro Woche')
plt.xticks(index + bar_width, label_week_lists)
plt.legend(bbox_to_anchor=(1, 1),
       bbox_transform=plt.gcf().transFigure)

def autolabel(rects, ax):
# Get y-axis height to calculate label position from.
    (y_bottom, y_top) = ax.get_ylim()
    y_height = y_top - y_bottom

    for rect in rects:
        height = rect.get_height()
        # Fraction of axis height taken up by this rectangle
        p_height = (height / y_height)
        # If we can fit the label above the column, do that;
        # otherwise, put it inside the column.
        if p_height > 0.95: # arbitrary; 95% looked good to me.
            label_position = height - (y_height * 0.05)
        else:
            label_position = height + (y_height * 0.01)

        ax.text(rect.get_x() + rect.get_width() / 2., label_position,
            '%d' % int(height),
            ha='center', va='bottom')

autolabel(rects1, ax)
autolabel(rects2, ax)
autolabel(rects3, ax)

plt.show()

Answer 1

If you print rect.get_height() values you get something like:

-8.159698443426123
11.395025597733763
-1.048913873322391
28.99318154295449
-7.552596893170488
7.959096278017519

Therefore height of a bar maybe a negative.

To mark negative bars modify if statement in auto label function as follow:

if p_height > 0.95: # arbitrary; 95% looked good to me.
    label_position = height - (y_height * 0.05) if (height > 0) else height + (y_height * 0.05) 
else:
    label_position = height + (y_height * 0.01) if (height > 0) else height - (y_height * 0.05) 

You have to ajust coefficients in else branch (0.05) manually because position of a below label depends on a font size of a label. (Position of a label for positive bar does not affected by font size because a label is above the bar).

Finale for my font settings:

Answer 2

mathplotlib is not very well documented in that department. Try using the dir() function to reveal the available options you have on the container you're working on. i found there is a .get_y() function which retuns negative numbers in that case

try this code instead

import pandas as pd
import matplotlib.pyplot as plt
from builtins import list
import matplotlib
matplotlib.style.use('ggplot')
import numpy as np

n_groups = 2

# create plot
fig, ax = plt.subplots()
fig.canvas.set_window_title('Mindestlohn Bundesweit')

index = np.arange(n_groups)
bar_width = 0.20
opacity = 0.8

list_reallohn_week_vollzeit = [-8.159698443426123, 11.395025597733763]
list_reallohn_week_teilzeit = [-1.048913873322391, 28.99318154295449]
list_reallohn_week_mini = [-7.552596893170488, 7.959096278017519]


rects1 = plt.bar(index + 0.00, list_reallohn_week_vollzeit, bar_width,
             alpha=opacity,
             color='b',
             label='Vollzeit')
rects2 = plt.bar(index + bar_width, list_reallohn_week_teilzeit, bar_width,
             alpha=opacity,
             color='g',
             label='Teilzeit')
rects3 = plt.bar(index + bar_width * 2,list_reallohn_week_mini, bar_width,
             alpha = opacity,
             color='c',
             label='Mini Job')

label_week_lists = ('2015 Nicht MdL berechtigt', '2015 mit MdL')

plt.ylabel('EUR')
plt.title('Reallöhne pro Woche')
plt.xticks(index + bar_width, label_week_lists)
plt.legend(bbox_to_anchor=(1, 1),
       bbox_transform=plt.gcf().transFigure)

def autolabel(rects, ax):
    # Get y-axis height to calculate label position from.
    (y_bottom, y_top) = ax.get_ylim()
    y_height = y_top - y_bottom

    for rect in rects:
    # print(dir(rect))

    height = 0
    if rect.get_y() < 0:
        height = rect.get_y()
    else:
        height = rect.get_height()

    print(rect.get_height())
    print( str(rect.get_y()) )

    # Fraction of axis height taken up by this rectangle
    p_height = (height / y_height)
    # If we can fit the label above the column, do that;
    # otherwise, put it inside the column.
    if p_height > 0.95:  # arbitrary; 95% looked good to me.
        label_position = height - (y_height * 0.05)
    else:
        label_position = height + (y_height * 0.01)

    ax.text(rect.get_x() + rect.get_width() / 2., label_position,
            '%d' % int(height),
            ha='center', va='bottom')

autolabel(rects1, ax)
autolabel(rects2, ax)
autolabel(rects3, ax)

plt.show()