如何将两列熊猫数据框绘制为水平条形图？

Question

I am trying to compare between 2 columns in my dataframe, and want to plot them as a horizontal barplot using Seaborn. In the plot below is just one column plotted ('budget2019').

1) How can I plot the second column ('budget2018') next to it, so that one can better see the evolution between both columns' values, and compare visually both bars, for each "ministry"?

2) How can I put the x-value at the end of each bar, so one can see (read) the x-value on each bar, and hasn't to go with the eyes down, and try to approximate the value? Because now on the actual plot one can't really read the x-value directly from a bar. (pretty difficult to get the exact value when looking at the plot)

3) After I get it to plot the 'budget2018' bars next to 'budget2019' bars, is there a way to also put the evolution in percent on the plot? So that it's visually better readable, how much evolution in % between both columns is?

Here is a look at my dataframe:

And here is the code I'm using to plot for now:

plt.figure(figsize=(15,8))

sns.set(style="darkgrid")

#ax = sns.barplot(x="budget2018", y="ministere", data=budget, label="Total")
sns.set_color_codes("pastel")
sns.barplot(x="budget2019", y="ministere", data=budget, label="Budget 2019")

sns.despine(left=True, bottom=True)

plt.tight_layout()
plt.show()

PS: If it is not doable with Seaborn, a solution with Matplotlib alone would be also okay. And what is needed, are horizonal bars, as in my plot, otherwise the yticks aren't readable.

EDIT (after using the code in ImportanceOfBeingErnest's comment):

Here is the plot I am actually seeing, after @ImportanceOfBeingErnest commented. It is pretty close to what I need.

Answer 1

To use seaborn, you have to use an intermediate dataframe created using melt to produce a "long-form" dataframe.

df2 = pd.melt(budget, id_vars=['ministere'], value_vars=['budget2018','budget2019'], var_name='year')

then plotting of both column is done using the hue= parameter:

fig,ax = plt.subplots()
sns.set_color_codes("pastel")
sns.barplot(x="value", y="ministere", hue="year", data=df2, palette='pastel')

adjusting the labels as requested is fairly easy (and you'll find plenty of other exaples on SO):

for i,m in budget.iterrows():
    ax.annotate(s='{:.2f}%'.format(m.loc['evolution_percent']),
                xy=(m.loc[['budget2018','budget2019']].max(),i),
                xytext=(10,0),
                textcoords='offset pixels',
                ha='left',
                va='center'
               )

labels = ['{:s}\n(2019: {:.2f}€)'.format(d.ministere,d.budget2019) for _,d in budget.iterrows()]
ax.set_yticklabels(labels)