def weightinginfo(feature1,feature2,im1,im2):
#weight=ee.mi(feature1,feature2)
'''X=feature1
Y=feature2
Xn=0
Xm=10
Yn=0
Ym=10
X=np.asarray(X)
Y=np.asarray(Y)
X=utils.quantise(X, 10, uniform='sampling', minmax=None, centers=False)
print X
Y=utils.quantise(Y, 10, uniform='sampling', minmax=None, centers=False)
sys = systems.DiscreteSystem(X,(Xn,Xm),Y,(Yn,Ym))
sys.calculate_entropies(method='qe',calc=['HX','HXY','HiXY','HshXY'])
weight=sys.I()'''
feature1=feature1.tolist()
feature2=feature2.tolist()
weight=ee.kldiv(feature1,feature2,k=3,base=2)
print 'weight is', weight
return weight
print('samples used', Ntry)
print('estimated MI', ent)
print('95% conf int.\n', err)
# DISCRETE ESTIMATORS
print("\n\nTest of the discrete entropy estimators\n")
print(
"For z = y xor x, w/x, y uniform random binary, we should get H(x)=H(y)=H(z) = 1, H(x:y) etc = 0, H(x:y|z) = 1"
)
x = [0, 0, 0, 0, 1, 1, 1, 1]
y = [0, 1, 0, 1, 0, 1, 0, 1]
z = [0, 1, 0, 1, 1, 0, 1, 0]
print("H(x), H(y), H(z)", ee.entropyd(x), ee.entropyd(y), ee.entropyd(z))
print("H(x:y), etc", ee.midd(x, y), ee.midd(z, y), ee.midd(x, z))
print("H(x:y|z), etc", ee.cmidd(x, y, z), ee.cmidd(z, y, x), ee.cmidd(x, z, y))
# KL Div estimator
print(
"\n\nKl divergence estimator (not symmetric, not required to have same num samples in each sample set"
)
print("should be 0 for same distribution")
sample1 = [[2 * random.random()] for i in range(200)]
sample2 = [[2 * random.random()] for i in range(300)]
print('result:', ee.kldiv(sample1, sample2))
print(
"should be infinite for totally disjoint distributions (but this estimator has an upper bound like log(dist) between disjoint prob. masses)"
)
sample2 = [[3 + 2 * random.random()] for i in range(300)]
print('result:', ee.kldiv(sample1, sample2))
ax[0].set_xlabel('Percent White')
plt.tight_layout()
fig.savefig('figures/MN_GA_density.pdf')
plt.close('all')
sns.set_style('white')
fig, ax = plt.subplots(1, 2, figsize=(6, 2))
# bins=np.arange(-.1,1.1,.05)
mn = EE.kldiv(EE.vectorize(np.random.normal(.82, np.sqrt(.82 * (1 - .82) / 100), 1000)), EE.vectorize(MN_values),
k=20)
ga = EE.kldiv(EE.vectorize(np.random.normal(.55, np.sqrt(.55 * (1 - .55) / 100), 1000)), EE.vectorize(GA_values),
k=20)
objects = [mn, ga]
y_pos = np.arange(len(objects))
ax[0].bar(y_pos * 2 + 1, objects, align='center', color=['darkturquoise', 'magenta'])
# plt.xticks(y_pos, objects)
# plt.ylabel('Usage')
# plt.title('Programming language usage')
ax[0].set_xticks(y_pos * 2 + 1)
ax[0].set_xticklabels(['Minnesota', 'Georgia'])
ax[0].set_xlim([0, 4])
ax[1].bar(y_pos * 2 + 1, [9.2, 13.3], align='center', color=['darkturquoise', 'magenta'])
# plt.xticks(y_pos, objects)
after_c = after["raw core"].values
before_xyz = []
before_xyzc = []
for j in range(len(before_x)):
before_xyz.append([before_x[j], before_y[j], before_z[j]])
before_xyzc.append(
[before_x[j], before_y[j], before_z[j], before_c[j]])
after_xyz = []
after_xyzc = []
for j in range(len(after_x)):
after_xyz.append([after_x[j], after_y[j], after_z[j]])
after_xyzc.append([after_x[j], after_y[j], after_z[j], after_c[j]])
kl_xyz, pw_xyz = ee.kldiv(before_xyz, after_xyz, return_pw=True)
plt.clf()
ax = fig.add_subplot(111, projection='3d')
cs = ax.scatter(before_x, before_y, before_z, c=pw_xyz, cmap=cm.jet)
plt.title("StudyID:" + meta_data["Study"][i] + "\nType:" +
meta_data['Type'][i] + "\nKL divergence= " + ("%.3f" % kl_xyz))
plt.colorbar(cs)
plt.savefig("figs/" + meta_data['Type'][i] + "_" + meta_data["Study"][i] +
".jpg",
bbox_inches='tight')
f_out = open(
"pointwise_kl/" + meta_data['Type'][i] + "_" + meta_data["Study"][i] +
".csv", 'w')
f_out.write("X,Y,Z,pointwise_kl_divergence\n")
tempent.sort() tempmean = np.mean(tempent) ent.append(tempmean) err.append((tempmean - tempent[samplo],tempent[samphi]-tempmean)) print 'samples used',Ntry print 'estimated MI',ent print '95% conf int.\n',err ## DISCRETE ESTIMATORS print "\n\nTest of the discrete entropy estimators\n" print "For z = y xor x, w/x,y uniform random binary, we should get H(x)=H(y)=H(z) = 1, H(x:y) etc = 0, H(x:y|z) = 1" x = [0,0,0,0,1,1,1,1] y = [0,1,0,1,0,1,0,1] z = [0,1,0,1,1,0,1,0] print "H(x),H(y),H(z)",ee.entropyd(x),ee.entropyd(y),ee.entropyd(z) print "H(x:y),etc",ee.midd(x,y),ee.midd(z,y),ee.midd(x,z) print "H(x:y|z),etc",ee.cmidd(x,y,z),ee.cmidd(z,y,x),ee.cmidd(x,z,y) ## KL Div estimator print "\n\nKl divergence estimator (not symmetric, not required to have same num samples in each sample set" print "should be 0 for same distribution" sample1 = [[2*random.random()] for i in range(200)] sample2 = [[2*random.random()] for i in range(300)] print 'result:',ee.kldiv(sample1,sample2) print "should be infinite for totally disjoint distributions (but this estimator has an upper bound like log(dist) between disjoint prob. masses)" sample2 = [[3+2*random.random()] for i in range(300)] print 'result:',ee.kldiv(sample1,sample2)
pass
else:
data_inter.append([val])
with open('../data/edges-by-type/{}_intra.dat'.format(prefix)) as ofile:
for line in ofile:
val = float(line)
if val == 0 or math.isnan(val):
pass
else:
data_intra.append([float(line)])
with open('../data/edges-by-type/{}_multi.dat'.format(prefix)) as ofile:
for line in ofile:
val = float(line)
if val == 0 or math.isnan(val):
pass
else:
data_multi.append([float(line)])
K = 5
KLxy = ee.kldiv(data_inter[:5000], data_intra[:5000], k = K)
KLyx = ee.kldiv(data_intra[:5000], data_inter[:5000], k = K)
JS = 0.5*(KLxy + KLyx)
print label_type, weight_type, JS