import numpy as np
from lib.timer import Timer
import breakout_detection
import runners.data_loader as data_loader
import matplotlib.pyplot as plt
SAMPLE_FILE_PATH = '../data/demo7.csv'
if __name__ == '__main__':
sw = Timer()
data = data_loader.load_data(SAMPLE_FILE_PATH)
sw.start()
edm_multi = breakout_detection.EdmMulti()
max_snp = max(max(data.values), 1)
# Z = [x/float(max_snp) for x in data.values]
Z = [x for x in data.values]
edm_multi.evaluate(Z, min_size=24, beta=0.001, degree=1)
print(sw.elapsed(f'data length: {len(data.values)}, using time:'))
plt.plot(np.asarray(data.index).tolist(), Z)
result = edm_multi.getLoc()
print(result)
for i in result:
plt.axvline(np.asarray(data.index).tolist()[i], color='#FF4E24')
# plt.plot(np.asarray(data.index).tolist()[i], np.asarray(data.values).tolist()[i], 'ro')
plt.show()
if i % (100) == 0:
dataSaver.add({
'ephoch': j,
'iter': i,
'totLoss': loss,
'fastConvergeEntropy': fcLoss,
'stableConvergeEntropy': stLoss,
'reward': reward
})
print(
'ephoc: ', j, '\titer: ', i, '\tloss: ',
roundDec(loss), '\treward: ', roundDec(reward),
'\ttimeElapsed: ',
timer.elapsed(step=(
i + j *
(mnist.train_size // config.batch_size))),
'\tfastConvergeEntropy :',
roundDec(fcLoss), '\tstableConvergeEntropy :',
roundDec(stLoss), '\tremaining: ', timer.left())
if j % (5) == 0:
print('Tot Time Elapsed: ', timer.elpasedTot(), ' after ',
j, ' steps')
if ((j % (25) == 0) & (j != 0)):
print(
'------------------ Saving Session ------------------')
saver.save(sess, modelSavePath)
print('------------------ Training Completed ------------------')
print('Tot Time Elapsed ', timer.elpasedTot())
