def fit(self, params):
if len(self.data) <= 1:
self.DEBUG_INFO = "No samples are there!"
self.changed("alert_generated")
return
X = np.asarray(self.data)[:, 0:2]
y = np.asarray(self.data)[:, 2]
self.clf = iSVC(C=params['C'],
gamma=params['gamma'],
degree=params['degree'],
coef0=params['coef0'],
kernel=params['kernel'])
self.clf.fit(X, y)
self.is_fitted = True
self.changed("model_fitted")
def fit(self, params):
if len(self.data) <= 1:
self.DEBUG_INFO = "No samples are there!"
self.changed("alert_generated")
return
X = np.asarray(self.data)[:, 0:2]
y = np.asarray(self.data)[:, 2]
self.clf = iSVC(
C=params["C"],
gamma=params["gamma"],
degree=params["degree"],
coef0=params["coef0"],
kernel=params["kernel"],
)
self.clf.fit(X, y)
self.is_fitted = True
self.changed("model_fitted")
# main function to test the SVC Model from matplotlib import pyplot as plt import utils from isvc import iSVC import numpy as np import isvc_gui as gui __author__ = 'morgan' clf = iSVC(display=True, gamma=1) # generate data X, y = utils.gen_noise_gauss(100) n = np.size(X, 0) big_x, big_y = utils.getBoxbyX(X, grid=30) big_xy = np.c_[big_x.reshape(big_x.size, 1), big_y.reshape(big_x.size, 1)] # add supervised labels y_semisupervised = np.zeros((n, 1)) n_perm = np.random.permutation(n) # random partition for labeling labeled_fraction = 0.1 num_labels = int(np.ceil(labeled_fraction*n)) # number of labeled examples based on the fraction of dataset size #print "Number of labels used: " + str(num_labels) y_semisupervised[n_perm[0:num_labels]] = y[n_perm[0:num_labels]].reshape(num_labels,1) # label selected examples #x_labeled = X[n_perm[0:num_labels], :] #y_labeled = y[n_perm[0:num_labels]] x_labeled_normal = X[np.where(y_semisupervised==-1)[0], :] x_labeled_anomaly = X[np.where(y_semisupervised==1)[0],:] x_labeled_unknown = X[np.where(y_semisupervised==0)[0],:] #build SVC model
__author__ = 'morgan'
# main function to test the SVC Model
from matplotlib import pyplot as plt
import utils
from isvc import iSVC
import numpy as np
import isvc_gui as gui
clf = iSVC(display=True, gamma=1)
#generate data
X, y = utils.gen_noise_gauss(100)
n = np.size(X, 0)
big_x, big_y = utils.getBoxbyX(X, grid=30)
big_xy = np.c_[big_x.reshape(big_x.size, 1), big_y.reshape(big_x.size, 1)]
# add supervised labels
y_semisupervised = np.zeros((n, 1))
n_perm = np.random.permutation(n) # random partition for labeling
labeled_fraction = 0.1
num_labels = int(np.ceil(
labeled_fraction *
n)) # number of labeled examples based on the fraction of dataset size
#print "Number of labels used: " + str(num_labels)
y_semisupervised[n_perm[0:num_labels]] = y[n_perm[0:num_labels]].reshape(
num_labels, 1) # label selected examples
#x_labeled = X[n_perm[0:num_labels], :]
#y_labeled = y[n_perm[0:num_labels]]
x_labeled_normal = X[np.where(y_semisupervised == -1)[0], :]
x_labeled_anomaly = X[np.where(y_semisupervised == 1)[0], :]
