In [1]:

    

%pylab inline


    

    




Populating the interactive namespace from numpy and matplotlib

In [2]:

    

import glob

tcs = glob.glob('ik-learning/tc/tc-*.npy')
print len(tcs), 'tcs found'


    

    




105 tcs found

In [3]:

    

tc = vstack([load(tc) for tc in tcs])
print 'mega shape', tc.shape


    

    




mega shape (105000, 3)

In [4]:

    

scatter(tc[:, 0], tc[:, 1])


    

    
Out[4]:





<matplotlib.collections.PathCollection at 0x10b7e00d0>






    

In [5]:

    

scatter(tc[:, 0], tc[:, 2])


    

    
Out[5]:





<matplotlib.collections.PathCollection at 0x10bc59450>






    

In [6]:

    

scatter(tc[:, 1], tc[:, 2])


    

    
Out[6]:





<matplotlib.collections.PathCollection at 0x10bf0de90>






    

In [7]:

    

from mpl_toolkits.mplot3d.axes3d import Axes3D

ax = axes(projection='3d')
ax.scatter(* tc.T)


    

    
Out[7]:





<mpl_toolkits.mplot3d.art3d.Path3DCollection at 0x10c1fab10>






    

In [8]:

    

from sklearn.cluster import KMeans

K = 150

kmeans = KMeans(K)
kmeans.fit(tc)


    

    
Out[8]:





KMeans(copy_x=True, init='k-means++', max_iter=300, n_clusters=150, n_init=10,
    n_jobs=1, precompute_distances=True, random_state=None, tol=0.0001,
    verbose=0)

In [9]:

    

ax = axes(projection='3d')
ax.scatter(* kmeans.cluster_centers_.T)


    

    
Out[9]:





<mpl_toolkits.mplot3d.art3d.Path3DCollection at 0x10fa44b10>






    

In [10]:

    

save('ik-learning/tc-{}.npy'.format(K), kmeans.cluster_centers_)


    

In [10]: