Seaborn Align twinx and x Axis-CodePudding

I am trying to align X axis with its twin but I'm not finding a way to do it.

Here is my code

# Initialize the figure
plt.figure(figsize=(16, 10))

# Adding a title
plt.title(f'Client Retention Quarters: Monthly Cohorts', fontsize = 14)

# Creating the heatmap
sns.heatmap(retention, annot = True,vmin = 0, vmax =30,cmap="flare", fmt='g')
plt.ylabel('Cohort Quarter')
plt.xlabel('')
plt.yticks( rotation='360')

#Twinx
ax2 = plt.twiny()
ax2.set_xticks(range(0,len(x2)))
ax2.set_xticklabels(labels=x2)
ax2.spines['top'].set_position(('axes', -0.10))
plt.show()

And here is the output:

I want to align the percentages with the x ticks.

Is it possible?

CodePudding user response：

You can use the below updated code. See if this works. Note that I have used random data for retention and x2. Basically, the main change it to get the xlim()s for both axes and then adjust it (see lambda f) so that the ticks align. Finally use set_major_locator() to fix the points. Hope this is what you are looking for...

retention = np.random.rand(10, 12) ##My random data
# Initialize the figure
plt.figure(figsize=(16, 10))

# Adding a title
plt.title(f'Client Retention Quarters: Monthly Cohorts', fontsize = 14)

# Creating the heatmap
ax=sns.heatmap(retention, annot = True,vmin = 0, vmax =30,cmap="flare", fmt='g') ## Note I am assigning to ax
plt.ylabel('Cohort Quarter')
plt.xlabel('')
plt.yticks( rotation='360')

x2 = np.around(np.linspace(1, 25, 12),2)
#Twinx
ax2 = ax.twiny()
#ax2.set_xticks(range(0,len(x2))) ## Commented as not required
#ax2.set_xticklabels(labels=x2) ## Commented as not required

## New code here ##
import matplotlib.ticker
l = ax.get_xlim()
l2 = ax2.get_xlim()
f = lambda y : l2[0] (y-l[0])/(l[1]-l[0])*(l2[1]-l2[0]) ##Add delta to each tick
ticks = f(ax.get_xticks())
ax2.xaxis.set_major_locator(matplotlib.ticker.FixedLocator(ticks)) ##Set the ticks

ax2.spines['top'].set_position(('axes', -0.10))
plt.show()