Bar Plot with inverted y axis and bars attached to bottom - python-3.x

Question

The code below creates a bar plot with an inverted y-axis. What I don't manage yet is that the bars do not "hang from above" but start at the bottom. In other words, I like the bars to start at the maximum value of the y axis (i.e. at the x-axis) and ending at the value of df['y']. How can I do that?
import pandas as pd
import matplotlib.pyplot as plt
df = pd.DataFrame(data={'x_cat': ['aaaaa',
'bvvvvvv',
'deeeee',
'qqqqqqq',
'rr rrrrrrrr',
'rss sdasr',
'cccccccccccc',
'aarrrrrrrrrrra'
],
'y': [11.91,
35.19,
43.61,
46.12,
75.03,
81.39,
83.28,
89.20]
})
df['rank'] = df['y'].rank(method='dense') - 1
fig = plt.figure()
ax = fig.add_subplot(111)
# increase space below subplot
fig.subplots_adjust(bottom=0.3)
ax.bar(df['rank'],
df['y'],
width=0.8,
)
# invert y axis
ax.invert_yaxis()
# label x axis
ax.set_xticks(range(len(df)))
ax.set_xticklabels(df['x_cat'],
fontdict={'fontsize': 14})
for tick in ax.get_xticklabels():
tick.set_rotation(90)

Answer 1

You would need to calculate the new bottom. (Note that
because the axis is inverted, the "bottom" becomes the visual top of the bars.) The bottom is the value, the height is maximum minus the value itself.
I changed some other aspects of your plot, e.g. if your values are not sorted, calculating the rank and using it for plotting would result in wrong labelling. Hence better sort the dataframe beforehands (and forget about the rank).
Finally, we would need to adjust the "sticky edges" of the bars, because they should sit tight to the bottom of the figure (i.e. the top of the axis).
import pandas as pd
import matplotlib.pyplot as plt
df = pd.DataFrame({'x_cat': ['aaaaa', 'bvvvvvv', 'deeeee', 'qqqqqqq', 'rr rrrrrrrr',
'rss sdasr', 'cccccccccccc', 'aarrrrrrrrrrra'],
'y': [11.91, 35.19, 43.61, 46.12, 75.03, 81.39, 83.28, 89.20]})
df.sort_values("y", inplace=True)
fig = plt.figure()
ax = fig.add_subplot(111)
# increase space below subplot
fig.subplots_adjust(bottom=0.3)
bars = ax.bar(df['x_cat'], df['y'].max()-df['y'], bottom=df['y'], width=0.8, )
# invert y axis
ax.invert_yaxis()
ax.tick_params(axis="x", rotation=90, labelsize=14)
for bar in bars:
bar.sticky_edges.y[:] = [df['y'].values.max()]
ax.autoscale()
plt.show()