Visualizing 2D data

Colormaps are used to map quantities stored at each point of a (discretized) surface to colours. Colours live in a 3D space. 2D data may contain metric or form information (or both). How can we choose a good colormap?

Perception
Intuition
Convention

By default, we want to use a sequential colormap (not assuming anything about the structure of our data).

References:

matplotlib: http://matplotlib.org/users/colormaps.html, https://matplotlib.org/users/dflt_style_changes.html
viscm: https://bids.github.io/colormap/
Another SciPy 2015 talk: https://www.youtube.com/watch?v=XjHzLUnHeM0



In [1]:

    
import pickle



In [2]:

    
data = pickle.load(open('data/correlation_map.pkl', 'rb'))



In [3]:

    
data.keys()









    Out[3]:





dict_keys(['emission energy', 'excitation energy', 'correlation'])



In [4]:

    
type(data['excitation energy'])









    Out[4]:





numpy.ndarray



In [5]:

    
data['excitation energy'].shape









    Out[5]:





(135,)



In [6]:

    
data['correlation'].shape









    Out[6]:





(55, 135)



In [7]:

    
import matplotlib
%matplotlib inline
matplotlib.style.use('ggplot')
import matplotlib.pyplot as plt



In [8]:

    
plt.imshow(data['correlation'])









    Out[8]:





<matplotlib.image.AxesImage at 0x7fbe87271860>

Are we dealing with form or metric information here?



In [9]:

    
matplotlib.__version__









    Out[9]:





'1.5.dev1'



In [10]:

    
from matplotlib.cm import magma, inferno, plasma, viridis
# from colormaps import magma, inferno, plasma, viridis



In [11]:

    
correlation = data['correlation']
excitation = data['excitation energy']
emission = data['emission energy']



In [12]:

    
plt.imshow(correlation,
           origin='lower',
           cmap=magma)









    Out[12]:





<matplotlib.image.AxesImage at 0x7fbe871a4550>



In [13]:

    
plt.imshow(correlation,
           origin='lower',
           extent=[excitation.min(), excitation.max(), emission.min(), emission.max()],
           cmap=magma)









    Out[13]:





<matplotlib.image.AxesImage at 0x7fbe8710b8d0>



In [14]:

    
f, ax = plt.subplots(1, 1, figsize=(5, 5))
map0 = ax.imshow(correlation,
          origin='lower',
          extent=[excitation.min(), excitation.max(), emission.min(), emission.max()],
          cmap=magma)
ax.plot(excitation, excitation)
ax.set_xlabel('Excitation Energy (eV)')
f.colorbar(map0, ax=ax)









    Out[14]:





<matplotlib.colorbar.Colorbar at 0x7fbe870446a0>

Revise and edit to best convey the scientific result.



In [15]:

    
f, ax = plt.subplots(1, 1, figsize=(5, 5))
ax.imshow(correlation,
          origin='lower',
          extent=[excitation.min(), excitation.max(), emission.min(), emission.max()],
          cmap=viridis)
ax.plot(excitation, excitation)
ax.set_xlabel('Excitation Energy (eV)')









    Out[15]:





<matplotlib.text.Text at 0x7fbe86fff470>

What about using the new default colormap (above)?