Ordering of categorical data labels from pandas dataframe in plot-CodePudding

I have data in the following format, where the column stage was created by df['stage'] = pd.Categorical(df['stage'], categories=['Undef', 'C', 'A', 'B'], ordered=True), to create an ordering as follows: Undef < C < A < B. When I print(df['stage']), this ordering is confirmed.

     stage                time
0    A     2012-01-26 21:52:56
1    A     2012-01-26 21:53:26
2    A     2012-01-26 21:53:56
3    A     2012-01-26 21:54:26
4    A     2012-01-26 21:54:56
..   ...                 ...
953  B     2012-01-27 05:49:26
954  B     2012-01-27 05:49:56
955  B     2012-01-27 05:50:26
956  C     2012-01-27 05:50:56
957  Undef 2012-01-27 05:51:26

This is how I plot the data:

fig, ax = plt.subplots()
ax.plot(df.time, df.stage)
plt.show()

The stages should be on the y axis and have the specified ordering, but they just appear in the ordering of their first appearance in the dataframe.

What am I doing wrong?

Edit: This is what the plot currently looks like:

All I want to do is to change the order of the y-labels to a custom order (and of course the corresponding plotted values).

Edit2: When adding the line df.sort_values(['stage'], inplace=True) this is what happens to the plot - which cannot be accurate as there can only be one stage at a time (although the ylabels are in the correct order now)

CodePudding user response：

Please find below the code to do what you are looking for. As the dates are in the x-axis, you will need to convert the labels to integers and back. So, first the dictionary is defined to show mapping of Undef, A, B, C as per your requirement. A list with the converted values is created and plot drawn. Finally, the updated y-axis ticklabels are done...

My data (input)

stage   time
0   A   2012-01-26 21:52:56
1   A   2012-01-26 21:53:26
2   A   2012-01-26 21:53:56
3   A   2012-01-26 21:54:26
4   A   2012-01-26 21:55:56
5   B   2012-01-26 21:56:56
6   B   2012-01-26 21:57:56
7   B   2012-01-26 21:58:56
8   B   2012-01-26 21:59:56
9   B   2012-01-26 22:00:56
10  C   2012-01-26 22:01:56
11  C   2012-01-26 22:02:56
12  C   2012-01-26 22:03:56
13  C   2012-01-26 22:04:56
14  Undef   2012-01-26 22:05:56
15  Undef   2012-01-26 22:06:56
16  Undef   2012-01-26 22:07:56

Plot without any sorting (your code equivalent)

Updated code

convert = {"Undef" : 0, "C" : 1, "A" : 2, "B" : 3}
stage_converted = []
for i in df.stage :
    stage_converted.append(conversion[i])

fig, ax = plt.subplots(figsize=(10, 5))
ax.plot(df.time, stage_converted)

### Update ticks and labels for y-axis
ax.set_yticks( list(convert.values()) )
ax.set_yticklabels( list(convert.keys()) )

plt.show()

...and finally the updated plot