Pandas Stacked Bar and Plot Issues

Question

I'm trying to overlay a stacked bar with a line plot. Running off pandas 0.25.3 working in vscode jupyter notebook. I have plotted this numerous times with different data sets using the relatively standard way below - but in this case I only get the latter matplot object plotted

fig, ax = plt.subplots()
ax2=ax.twinx()
df1.plot(kind='bar',stacked=True, ax=ax)
df2.plot(ax=ax2,linestyle='-', marker='o')

df1 and df2 by design have the same index. I have also tried merging these in case that was the driving issue as well as combinations using secondary_y, setting ax2 object to the matplot object. I tried the below work around too and tried simplifying to a bar chart on the primary axis. Secondary and primary line graphs plot fine too, so no issue with the underlying data (nan free etc)

pandas DataFrame how to mix bar and line plots with different scales

Data is the form as below (when joined) - where col3 is the line plot, col1 and col2 are the stacked bars

idx   col1    col2    col3
1.0    0.2    0.8    0.6
2.0    0.1    0.9    0.52
3.0    0.1    0.9    0.5
4.0    0.05    0.95    0.43

There are numerous questions asking similar on this site, but they tend to be more for beginners. Am I missing a version compatibility issue? I've written this plot before in 30secs but have spent too much time trying to figure out the issue here. Has anyone seen this before?

Is this a case of https://stackoverflow.com/questions/46645280/pyplot-move-alternative-y-axis-to-background — BigBen, May 21 '20 at 01:38
I did see that too and doesn't seem to be the solution. Slightly pulling my hair out here. — user2065472, May 21 '20 at 02:14
Does the sample data represent your actual data? Specifically, is your data time-indexed? If not, how many rows are there in your data? — Quang Hoang, May 21 '20 at 02:19
The data is representative - 8 rows, 13 columns. Index is float. First 12 columns sum to 1 (100%), with the final column a decaying function <1, > 0. — user2065472, May 21 '20 at 11:17
Update here - I ended up getting this to plot by setting the index to type str! — user2065472, May 21 '20 at 11:59
@QuangHoang - I appreciate the questioning above - it got me thinking about the datatype of the index.... — user2065472, May 21 '20 at 12:00

user2065472 · Accepted Answer · 2020-05-21T16:16:50.723

0

A simple resetting of the datatype of the index did the trick for me here:

df1.index = df1.index.astype(str)
df2.index = df2.index.astype(str)

There is some loss of functionality as I had been plotting a vline between ticks, but this appears to be the best solution I could find.

edited May 21 '20 at 16:16

answered May 21 '20 at 15:24

user2065472

3
3

Pandas Stacked Bar and Plot Issues

1 Answers1