def calculate_dist(data):
"""
This function replaces NaN values for median and scaling values to [0 1]
it also calculates three vectors of Euclidean,
Mahalanobis and Cosine distances between centroid of rows and
each row in dataframe
Returns:
three vectors of distances
"""
import scipy.spatial.distance as dist
import numpy as np
import scipy as sp
from ploting import boxplot, histogram
# scaling
del data['Patient_id']
for k in data.columns:
data[k] = data[k].fillna(round(data[k].median()))
data[k] = (data[k] - data[k].min())
data[k] = (data[k] - data[k].min()) / data[k].max()
if data[k].isnull().sum() > len(data[k]) * 0.8 or len(set(
data[k])) == 1:
del data[k]
# Calculate centroid
centroid = np.mean(data)
centroid = centroid.as_matrix()
numpyMatrix = data.as_matrix()
#Calculate covariance matrix
covmx = data.cov()
invcovmx = sp.linalg.pinv(covmx)
#Calculate Euclidean,Mahalanobis and Cosine distance
Mahaldist, Eucliddist, Cosinedist = [], [], []
for h in range(len(numpyMatrix)):
Mahaldist.append(dist.mahalanobis(numpyMatrix[h], centroid, invcovmx))
Eucliddist.append(dist.euclidean(numpyMatrix[h], centroid))
Cosinedist.append(dist.cosine(numpyMatrix[h], centroid))
# ploting
boxplot(Eucliddist, Mahaldist, Cosinedist)
histogram(Eucliddist, Mahaldist, Cosinedist, len(data))
return Mahaldist, Eucliddist, Cosinedist
plt.plot(r, r * model_m.T[k], 'r', alpha=0.5)
if n != 1:
plt.ylabel(r'$\Delta \bar z_{' + str(k + 1) + '} r$', fontsize=14)
else:
plt.ylabel(r'$\Delta \bar z r$', fontsize=14)
plt.axhline(y=0.0, xmin=0.0, xmax=1.0, linewidth=1.5, color='k')
plt.xscale('log')
if (k + 1) % row == 0 or k + 1 == n:
plt.xlabel(r'$r$ (Mpc)')
plt.subplots_adjust(hspace=0.26,
wspace=0.56,
right=0.95,
left=0.16,
top=0.93,
bottom=0.10)
plt.suptitle('Delta Average Redshift*Radius vs. Radius')
plt.show()
'''
ploting.histogram(b_g_rel[2],100,range=(0.,3.),color='r',xlabel='Redshift',ylabel='counts',title='Log of Background Galaxies Histogram',yscale='log')
ploting.plot_posit(b_g_rel[0],b_g_rel[1],color='r')
'''
program_end = time.time()
run_time = program_end - program_start
print("done")
print("total run time: " + str(run_time) + " sec")
print(" or " + str(run_time / 60.) + " min")
print(" or " + str(run_time / 3600.) + " hr")