test.py

import matplotlib.pyplot as plt
import numpy as np
from scipy.stats import multivariate_normal
import random as rn
import iso
mean = [0, 0]
cov = [[1, 0], [0, 1]]  # diagonal covariance
Nobjs = 2000
x, y = np.random.multivariate_normal(mean, cov, Nobjs).T
x[0]=3
y[0]=3
X=np.array([x,y]).T

ntrees=500
CT=[]
sample = 256
for i in range(ntrees):
    ix = rn.sample(range(Nobjs),sample)
    X_p = X[ix]
    limit = int(np.ceil(np.log2(sample)))
    C=iso.iTree(X_p,0,limit)
    CT.append(C)


S = np.zeros(Nobjs)
c = iso.c_factor(sample)
for i in range(Nobjs):
    h_temp = 0
    for j in range(ntrees):
        h_temp += iso.PathFactor(X[i],CT[j]).path*1.0
    Eh = h_temp/ntrees
    S[i] = 2.0**(-Eh/c)


ss=np.argsort(S)
plt.plot(x,y,'bo')
plt.plot(x[ss[-10:]],y[ss[-10:]],'ro')

plt.figure()

sv1 = []
sv2 = []
sv3 = []

for j in range(ntrees):
    sv1.append(2**(-iso.PathFactor(X[ss[0]],CT[j]).path*1.0/c))
    sv2.append(2**(-iso.PathFactor(X[ss[Nobjs/2]],CT[j]).path*1.0/c))
    sv3.append(2**(-iso.PathFactor(X[ss[-1]],CT[j]).path*1.0/c))

plt.plot(sv1,label='normal')
plt.plot(sv2, label='semi')
plt.plot(sv3, label='outlier')
plt.legend(loc=0)

plt.show()