Change main plot legend label text

Question

So far I have been able to label the subplots just fine but I'm having an issue with the main one.

Here's the relevant part of my code:

data_BS_P = data[channels[0]]
data_BS_R = data[channels[1]]
data_BS_Y = data[channels[2]]
plot_BS_P = data_BS_P.plot() #data_BS_P is a pandas dataframe
axBS = plot_BS_P.gca()
axBS.plot(data_BS_R, label='Roll')
axBS.plot(data_BS_Y, label='Yaw')
axBS.set_ylabel('Amplitude (urad)')
axBS.legend(loc='upper center', bbox_to_anchor=(0.5, 1.05), ncol=3,
            fancybox=True, shadow=True)
ml1 = MultipleLocator(10)
ml2 = MultipleLocator(3600)
axBS.yaxis.set_minor_locator(ml1)
axBS.xaxis.set_minor_locator(ml2)
plot_BS_P.save('L1-SUS-BS_M1_DAMP_PRY_INMON.jpg')

And this is what I have so far: Notice the lengthy label for the blue line. I'd like that to be labeled as "Pitch" instead of the file name. In which line can I do that?

Delaryd

You haven't posted enough code to solve this problem. Post the code where the blue line is plotted. I assume the problem is in this line plot_BS_P = data_BS_P.plot(). – Cheesecloth

@ebarr, plot_BS_P = data_BS_P.plot() did it. data_BS_P is likely a pandas DataFrame, which has it own name blot-in. Am I right, O.P.? – Bestir

It's the data_BS_P.plot() that plots the blue line. The line above this is data = TimeSeriesDict.fetch(channels, start, end, verbose=True) and the rest is the massive list of channels I have to download... – Delaryd

Answer 1

You need to gain access of the legend() object and use set_text() to change the text values, a simple example:

plt.plot(range(10), label='Some very long label')
plt.plot(range(1,11), label='Short label')
L=plt.legend()
L.get_texts()[0].set_text('make it short')
plt.savefig('temp.png')

In your case, you are changing the first item in the legend, I am quite sure the 0 index in L.get_texts()[0] applies to your problem too.

Bestir

In interactive mode, add plt.gcf().canvas.draw() to redraw the figure with the new label. (In the provided example, plt.savefig updates the figure before it is saved.) – Gilding

a bit shorter : L.texts[0]..set_text('shorter') – Gratian

Can't get this to work with xarray.DataArray.plot.line() – Excelsior

Answer 2

Another way:

ax.legend(labels=mylabels)

Cuellar

This was by far the easiest way presented here for my use case. Thank you. – Brazell

equivalently, if ax isn't explicitly defined then of course plt.legend(labels=my_labels) where my_labels is a list. – Lowboy

This broke the legend patch colors when used on boxplots made with seaborn. – Zobe

I used it but it does not preserve the colors I don't know why... – Kathlenekathlin

how to pass a dictionary ? seems not to work, i don't how which is the order of classes... – Tees

It provides the proper labels, but colors instead completely disappeared. – Jeremie

Answer 3

To avoid messing up color and marker size of the legend keys when plotting with seaborn:

handles, previous_labels = ax.get_legend_handles_labels()
ax.legend(handles=handles, labels=new_labels)

Duer

Answer 4

The answer by ksindi works for setting the labels, but as some others commented, it can break the legend colours when used with seaborn (in my case a scatterplot: the dots and text didn't line up properly anymore). To solve this, also pass the handles to ax.legend.

# the legend has often numbers like '0.450000007', the following snippet turns those in '0.45'
label_list = []
for t in ax.get_legend_handles_labels():
    # the first result will be all handles, i.e. the dots in the legend
    # the second result will be all legend text
    label_list.append(t)

new_list = []
for txt in label_list[1]:
    if txt[0] == '0':
        txt = str(txt)[:4]
    new_list.append(txt)
label_list[1] = new_list

ax.legend(handles=label_list[0], labels=label_list[1])

(I would have posted this as a comment, but don't have enough reputation yet)

Lan