def __init__(self, num_try, num_labels, labels_X_t, labels_X_t_k_m,
labels_Y_t_k_m):
# self.num_try = num_try #TODO rename
self.labels_surrogateaverage = []
self.labels_surrogatediv = []
for label_i in range(num_labels):
temp_causalities = []
for i in range(num_try):
shuffled_X_t = labels_X_t[label_i][:]
shuffle(shuffled_X_t)
shuffled_X_t_k_m = labels_X_t_k_m[label_i][:]
shuffle(shuffled_X_t_k_m)
shuffled_Y_t_k_m = labels_Y_t_k_m[label_i][:]
shuffle(shuffled_Y_t_k_m)
causality_calculatorYtoX = CausalityCalculator(
shuffled_X_t, shuffled_X_t_k_m, shuffled_Y_t_k_m)
temp_causalities.append(
causality_calculatorYtoX.calcGrangerCausality(
)) #TODO TODO False が0として扱われているのでキケン
self.labels_surrogateaverage.append(np.average(temp_causalities))
self.labels_surrogatediv.append(np.std(temp_causalities))
def __init__(self, num_try, num_labels,
labels_X_t, labels_X_t_k_m, labels_Y_t_k_m):
self.labels_surrogateaverage = []
self.labels_surrogatediv = []
for label_i in range(num_labels):
temp_causalities = []
for i in range(num_try):
shuffled_X_t = labels_X_t[label_i][:]
shuffle(shuffled_X_t)
shuffled_X_t_k_m = labels_X_t_k_m[label_i][:]
shuffle(shuffled_X_t_k_m)
shuffled_Y_t_k_m = labels_Y_t_k_m[label_i][:]
shuffle(shuffled_Y_t_k_m)
causality_calculatorYtoX = CausalityCalculator(shuffled_X_t, shuffled_X_t_k_m, shuffled_Y_t_k_m)
# TODO False が0として扱われているのでキケン
temp_causalities.append(causality_calculatorYtoX.calcRegularizedGrangerCausality(0.99, 0.0001, 0.0001, 0.0001))
self.labels_surrogateaverage.append(np.average(temp_causalities))
self.labels_surrogatediv.append(np.std(temp_causalities))
labels_pcaed_Xt.append(pca.applyPCA(labels_frames_embeddedVecXt[i]))
labels_pcaed_Xtkm.append(pca.applyPCA(labels_frames_embeddedVecXtkm[i]))
labels_pcaed_Ytkm.append(pca.applyPCA(labels_frames_embeddedVecYtkm[i]))
# calc Causalities
from CausalityCalculator import CausalityCalculator
# from SurrogatingTester import SurrogatingTester
# num_try = 100
# tester = SurrogatingTester(num_try, numCluster, labels_frames_embeddedVecXt, labels_frames_embeddedVecXtkm, labels_frames_embeddedVecYtkm)
causalities = []
for k in range(numCluster):
print np.array(labels_frames_embeddedVecXt[k]).shape
calclator = CausalityCalculator(labels_frames_embeddedVecXt[k],
# 2番めの引数の影響を除く
labels_frames_embeddedVecXtkm[k],
labels_frames_embeddedVecYtkm[k])
causalities.append(calclator.calcRegularizedGrangerCausality(0.99, 0.0001, 0.0001, 0.0001))
# surrogated_causalities = tester.compare(causalities)
# print tester.get()
# frames_vals = [surrogated_causalities[label_i] for frame_i, label_i in enumerate(frames_labels)]
frames_vals = [causalities[label_i] for frame_i, label_i in enumerate(frames_labels)]
# frames_vals = [eigen_vals[label_i][0] for frame_i, label_i in enumerate(frames_labels)]
maker = MovieMaker(num_canvas=2, num_canvas_horizontal=2)
maker.addSomeSticksMovie(frames_positions_x, [[0, 1, 2, 3], [3, 6, 5, 4], [2, 6]], 1, 1, "X")
maker.addSomeSticksMovie(frames_positions_y, [[0, 1, 2, 3], [3, 6, 5, 4], [2, 6]], 1, 1, "Y")
# maker.addPCAMovie(range(len(frames_positions_x)),
# frames_vals, pcaed_data, 1, 1, frames_labels=frames_labels)
maker.savefigs(saveDir, 98)
from CausalityCalculator import CausalityCalculator
import numpy as np
# audio = np.transpose(np.loadtxt("/Users/kawano/Desktop/audio.csv", delimiter=","))
# video = np.transpose(np.loadtxt("/Users/kawano/Desktop/video.csv", delimiter=","))
# symptoms = np.transpose(np.loadtxt("/Users/kawano/Desktop/symptoms.csv", delimiter=","))
audio = np.loadtxt("/Users/kawano/Desktop/audio.csv", delimiter=",")
video = np.loadtxt("/Users/kawano/Desktop/video.csv", delimiter=",")
symptoms = np.loadtxt("/Users/kawano/Desktop/symptoms.csv", delimiter=",")
print audio
calc = CausalityCalculator(video, audio, symptoms)
calc = CausalityCalculator(audio, video, symptoms)
eigen_val, eigen_vec = calc.calcRegularizedGrangerCausality(0.99, 0.0, 0.0, 0.0)
print eigen_val
print "pcca = %f" % eigen_val ** 0.5
