karenyyng/matplotlib_cheatsheet.md

## matplotlib_cheatsheet.md

      
    Raw
  

              matplotlib_cheatsheet.md
            
          
    All things about contour plot

Contour plot with log axes

Post
# approach 1 
z = np.arange(100).reshape((10,10)) 
x = np.logspace(0, 4, 10) 
y = np.logspace(0, 4, 10) 

ax1 = subplot(121) 
ax1.contour(np.log10(x), np.log10(y), z) 

# approach 2 
ax2 = subplot(122) 
ax2.set_xscale("log") 
ax2.set_yscale("log") 
ax2.contour(x, y, z) 
Making contour plot or so with log colors

ref
from matplotlib.colors import LogNorm
plt.pcolor(X, Y, Z1, norm=LogNorm(vmin=Z1.min(), vmax=Z1.max()), cmap='PuBu_r')
similarly specifying color range for contour plot

Use vmin and vmax keyword.
putting color bar on a different axis

SO post
ig, axes = plt.subplots(nrows=2, ncols=2)
for ax in axes.flat:
    im = ax.imshow(np.random.random((10,10)), vmin=0, vmax=1)

fig.subplots_adjust(right=0.8)
cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])
fig.colorbar(im, cax=cbar_ax)

Plot aspects

plt.axes().set_aspect('equal')   # ratio of long to short dimensions
when you want the X and Y axes have the same size dimensions for each unit.
making plots with multiple panels (subplots)

perhaps the best approach

using gridspec
import matplotlib.gridspec as gridspec
# specify the overall figure size 
plt.figure(figsize=(6, 6))
# specify a 2 x 2 grid 
gs = gridspec.GridSpec(2, 2, width_ratios=[1, 1], 
                       height_ratios=[1, 1])
gs.update(hspace=0.2, wspace=0.5)
# add a lower triangle and have each axis named accordingly 
ax00 = plt.subplot(gs[0, 0])
ax10 = plt.subplot(gs[1, 0])
ax11 = plt.subplot(gs[1, 1])
# use each axis to make whatever plots!
More complicated approach


with more fine grained control

fig = plt.figure()
ax1 = fig.add_subplot(<row_no>, <col_no>, <plot_no_1>)
ax1.plot()

ax2 = fig.add_subplot(<row_no>, <col_no>, <plot_no_2>)
ax2.plot()
...
# saving all the subplots
fig.savefig("plot_name.png", bbox_inches='tight')
slightly less complicated approach

ref
Colormap

ref
Revert colormap

append _r to the name of the colormap, e.g. use cubehelix_r instead of cubehelix
Rotate axis labels

plt.xticks(rotation=45)


SO post
Or
ref
locs, labels = plt.xticks()

plt.setp(labels, rotation=90)
plt.plot(x, delay)
Adjust subplots spacing

SO post
colorbar label size

cbar = fig.colorbar(surf, shrink=0.5, aspect=20, fraction=.12,pad=.02)
cbar.set_label('Activation',size=18)
# access to cbar tick labels:
cbar.ax.tick_params(labelsize=5) 
ref
plot frame

remove upper / right axis line
[remove upper / right axis ticks]

ax.tick_params(labeltop='off', labelright='off')
ax.yaxis.set_ticks_position('left')
ax.xaxis.set_ticks_position('bottom')
[change the direction of the ticks]

mpl.rcParams['xtick.direction']= out
mpl.rcParmas['ytick.direction'] = out
color blind friendly map

ref
Fix in ~/matplotlib/matplotlib.rc
# colorblind friendly color cycler 
axes.prop_cycle    : cycler('color', 'E69F00, 56B4E9, 009E73, 0072B2, D55E00, CC79A7, F0E442, 000000')
Alternative colormap tableau colors
pandas scattermatrix plotting

control axis limits for Pandas kde plots
from pandas.plotting import scatter_matrix
axes = scatter_matrix(pointOfSalesDf[:], alpha=0.1, 
                      figsize=(10, 10), diagonal='kde')
_ = axes[0][0].set_xlim(*sales_xlim)
_ = axes[0][0].set_ylim(*inv_xlim)

_ = axes[1][0].set_xlim(*sales_xlim)
_ = axes[1][0].set_ylim(*inv_xlim)

_ = axes[0][1].set_xlim(*inv_xlim)
_ = axes[0][1].set_ylim(*sales_xlim)

_ = axes[1][1].set_xlim(*inv_xlim)
_ = axes[1][1].set_ylim(*sales_xlim)
pandas KDE

ax = pointOfSalesDf.OHInvUnts_WTD.plot.kde()
_ = ax.set_xlim(-1e4, 5e3)

legend

plt.plot(x, y, title='STUFF')
plt.legend()
legend outside plot

ref
use bbox_to_anchor keyword
plt.legend(loc="upper left", bbox_to_anchor=(1,1))