I am sharing Y-axis in two subplots, with the following codes but both shared plots are missing legends in them.
projectDir = r'/media/DATA/banikr_D_drive/model/2021-04-28-01-18-15_5_fold_114sub'
logPath = os.path.join(projectDir, '2021-04-28-01-18-15_fold_1_Mixed_loss_dice.bin')
with open(logPath, 'rb') as pfile:
h = pickle.load(pfile)
print(h.keys())
fig, ax = plt.subplots(2, figsize=(20, 20), dpi=100)
ax[0].plot(h['dice_sub_train'], color='tab:cyan', linewidth=2.0, label="Train")
ax[0].plot(smooth_curve(h['dice_sub_train']), color='tab:purple')
ax[0].set_xlabel('Epoch/iterations', fontsize=20)
ax[0].set_ylabel('Dice Score', fontsize=20)
ax[0].legend(loc='lower right', fontsize=20)#, frameon=False)
ax1 = ax[0].twiny()
ax1.plot(h['dice_sub_valid'], color='tab:orange', linewidth=2.0, alpha=0.9, label="Validation" )
ax1.plot(smooth_curve(h['dice_sub_valid']), color='tab:red')
# , bbox_to_anchor = (0.816, 0.85)
ax[1].plot(h['loss_sub_train'], color='tab:cyan', linewidth=2.0, label="Train")
ax[1].plot(smooth_curve(h['loss_sub_train']), color='tab:purple')
ax2 = ax[1].twiny()
ax2.plot(h['loss_sub_valid'], color='tab:orange', linewidth=2.0, label="Validation", alpha=0.6)
ax2.plot(smooth_curve(h['loss_sub_valid']), color='tab:red')
ax[1].set_xlabel('Epoch/iterations', fontsize=20)
ax[1].set_ylabel('loss(a.u.)', fontsize=20)
ax[1].legend(loc='upper right', fontsize=20)
# ,bbox_to_anchor = (0.8, 0.9)
plt.suptitle('Subject wise dice score and loss', fontsize=30)
plt.setp(ax[0].get_xticklabels(), fontsize=20, fontweight="normal", horizontalalignment="center") #fontweight="bold"
plt.setp(ax[0].get_yticklabels(), fontsize=20, fontweight='normal', horizontalalignment="right")
plt.setp(ax[1].get_xticklabels(), fontsize=20, fontweight="normal", horizontalalignment="center")
plt.setp(ax[1].get_yticklabels(), fontsize=20, fontweight="normal", horizontalalignment="right")
plt.show()
Any idea how to solve the issue? [1]: https://i.sstatic.net/kg7PY.png
ax1 has a twin y-axis with ax[0], but they are two separate axes. That's why ax[0].legend() does not know about the Validation line of ax1.
To have Train and Validation on the same legend, plot empty lines on the main axes ax[0] and ax[1] with the desired color and label. This will generate dummy Validation entries on the main legend:
...
ax[0].plot([], [], color='tab:orange', label="Validation")
ax[0].legend(loc='lower right', fontsize=20)
...
ax[1].plot([], [], color='tab:orange', label="Validation")
ax[1].legend(loc='upper right', fontsize=20)
...