I have a data frame that looks something like this:

product version count_before    count_after
AdM     0      770422.0         449396.0
AdM     2      732007.0         57480.0
AdM     5      NaN  477056.0
AdM     1      1071.0   309.0
AllT    0      14.0 NaN
... ... ... ...
Zam     1      973.0    415.0
Zam     0      6682982.0    465034.0
leg     0      12741.0  5573.0
leg     2      12031.0  918.0
leg     5      NaN  8794.0

Shape (105, 3)

I would like to plot for each company a grouped barplot that shows the count_before and count_after if that is possible. I’ve tried plotting directly from a dataframe or using matplotlib but I couldn’t make the plot easy to read.

Now, I’m thinking of plotting each product separately in subplots so that in each subplot I have essentially have the same product plotting twice, once with count_before with version 0,1,2,5 and with count_after with the same.

As an example I managed to create a lineplot in subplots a while ago for each product using some other metric

I’d like to create something similar but using a barplot.

So far I have tried using seaborn to plot it but couldn’t manage it.

I’ve also tried creating a pivot table out of the dataset so I ended up having a df pivoted with count_before then one with count_after.

version      0  1   2
publisher           
AdM         770422.0    1071.0  732007.0
All         14.0    0.0 14.0
.........................

Any ideas on how I could achieve this?