Seaborn因子图的对数网格线

Question

I am trying to plot a logarithmic plot using seaborn factorplot on a dataframe as follows

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

l1 = [0.476, 0.4427, 0.378, 0.2448, 0.13, 0.004, 0.012, 0.0933, 3.704e-05, 
    1.4762e-06, 4.046e-08, 2.99e-10, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]

df = pd.DataFrame(l1, columns=["y"])
df.reset_index(inplace=True)

g = sns.factorplot(x='index',y='y', data=df, aspect=2, size=8)
g.fig.get_axes()[0].set_yscale('log')
plt.grid(True,which="both",ls="--",c='gray')  

plt.show()

I get the following figure.

Even though I changed the Y-axis scale to log and used both the gridlines, the final figure doesnt have the log scale ticks. However, the same code when used with another set of values gives me the following figure. In this case, the minimum value is limited to 10^-7

l2 = [0.29, 0.111, 0.0285, 0.0091, 0.00045, 5.49759e-05, 1.88819e-06, 0.0, 0.0, 0.0]
df = pd.DataFrame(l2, columns=["y"])
# same code as above

Any idea where I am being wrong?

---

Update 1

I follwed Diziet's answer and forced the major and minor ticks as follows

 g.ax.yaxis.set_minor_locator(tkr.LogLocator(base=10, subs='all')) g.ax.yaxis.set_minor_formatter(tkr.NullFormatter()) g.ax.set_yscale('log') g.ax.grid(True,which="both",ls="--",c='gray') 

But it still doesnt solve the problem

Answer 1

The problem is that in order to set the locations of ticks for cases where the automatically chosen major ticks are more than a decade away from each other seems to require to set the subs parameter of the locator as well as the numticks manually. Here, mticker.LogLocator(base=10, subs=np.arange(0.1,1,0.1),numticks=10) .

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.ticker as mticker
import seaborn as sns

l1 = [0.476, 0.4427, 0.378, 0.2448, 0.13, 0.004, 0.012, 0.0933, 3.704e-05, 
    1.4762e-06, 4.046e-08, 2.99e-10, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] 
df = pd.DataFrame(l1, columns=["y"])
df.reset_index(inplace=True)

g = sns.factorplot(x='index',y='y', data=df, aspect=2, size=8)
g.ax.set_yscale('log')

plt.grid(True,which="both",ls="--",c='gray')  

locmin = mticker.LogLocator(base=10, subs=np.arange(0.1,1,0.1),numticks=10)  
g.ax.yaxis.set_minor_locator(locmin)
g.ax.yaxis.set_minor_formatter(mticker.NullFormatter())

plt.show()

More generally also look at this question .

Answer 2

I do not think you are doing anything wrong. It seems to me that, in your first example, matplotlib decided (for a reason unknown to me) not to show any minor ticks, while it does for the second example.

One way to solve your issue is to force the display of minor ticks:

g = sns.factorplot(...)

ax = g.axes.flatten()[0]
ax.set_yscale('log')
ax.yaxis.set_minor_locator(matplotlib.ticker.LogLocator(base=10.0, subs='all'))
ax.yaxis.set_minor_formatter(matplotlib.ticker.NullFormatter())

plt.grid(True,which="both",ls="--",c='gray')