如何在 matplotlib 中创建堆叠条形图，其中堆栈因条形而异？

Question

So I have a pandas DataFrame that looks something like this:

       year country  total
0  2010     USA     10
1  2010    CHIN     12
2  2011     USA      8
3  2011    JAPN     12
4  2012    KORR      7
5  2012     USA     10
6  2013    CHIN      9
7  2013     USA     13

I'd like to create a stacked bar chart in matplotlib, where there is one bar for each year and stacks for the two countries in that year with height based on the total column. The color should be based on the country and be represented in the legend.

I can't seem to figure out how to make this happen. I think I could do it using for loops to go through each year and each country, then construct the bar with the color corresponding to values in a dictionary. However, this will create individual legend entries for each individual bar such that there are 8 total values in the legend. This is also a horribly inefficient way to graph in matplotlib as far as I can tell.

Can anyone give some pointers?

Answer 1

You need to transform your df first. It can be done via the below:

df = pd.DataFrame({'year': {0: 2010, 1: 2010, 2: 2011, 3: 2011, 4: 2012, 5: 2012, 6: 2013, 7: 2013},
                   'country': {0: 'USA', 1: 'CHIN', 2: 'USA', 3: 'JAPN', 4: 'KORR', 5: 'USA', 6: 'CHIN', 7: 'USA'},
                   'total': {0: 10, 1: 12, 2: 8, 3: 12, 4: 7, 5: 10, 6: 9, 7: 13}})

df2 = df.groupby(['year',"country"])['total'].sum().unstack("country")
print (df2)

#
country  CHIN  JAPN  KORR   USA
year                           
2010     12.0   NaN   NaN  10.0
2011      NaN  12.0   NaN   8.0
2012      NaN   NaN   7.0  10.0
2013      9.0   NaN   NaN  13.0
#

ax = df2.plot(kind='bar', stacked=True)

plt.show()

Result: