def quickPlot(lnp, sampler, p0, p1, p2):
""" Generates a figure showing the data, the best fit and 3 sigmas
intervals (if possible), and the residuals """
left = 0.125
width = 1.0 - 2 * 0.125
rect1 = [left, 0.34, width, 1.0 - 0.1 - 0.34]
rect2 = [left, 0.1, width, 0.34 - 0.1]
fig = figure.figure()
ax0 = fig.add_axes(rect1)
ax1 = fig.add_axes(rect2, sharex=ax0)
plotData(lnp, ax=ax0, lw=2.0, label="Data", zorder=5, color="#ff0000")
plotFit(lnp, p0, p1, p2, ax=ax0, zorder=0, lw=1, color="0.0")
# plotCI(lnp, sampler, ax=ax0, zorder=-5, alpha=0.3)
ax0.set_xlabel("X")
ax0.set_ylabel("Y")
l = ax0.legend(numpoints=1, scatterpoints=1)
l.draw_frame(False)
l.draggable(True)
figure.setp(ax0.get_xticklabels(), visible=False)
figure.theme(ax=ax0)
ax0.yaxis.set_major_locator(figure.MaxNLocator(5, prune="lower"))
plotResiduals(lnp, sampler, p0, p1, p2, ax=ax1)
ax1.set_xlabel("X")
ax1.set_ylabel("Residuals")
figure.theme(ax=ax1)
ax0.yaxis.set_major_locator(figure.MaxNLocator(4, prune="both"))
return ax0, ax1
def __init__(self, amount = 1):
self.screen = [[' '] * ( wight) for i in range(height)]
self.active = False
self.is_hard = False
if amount >= 2:
if amount > 2:
self.is_hard = True
self.active = True
self.curr2 = figure()
self.curr_color2 = 0
self.level2 = 0
self.killed2 = 0
self.scores2 = 0
self.screen2 = [[' '] * wight for i in range(height)]
self.delete2 = []
self.scores = 0
self.level = 0
self.killed = 0
self.curr = figure()
self.curr_color = 0
self.delete = []
def rotate(self, arg, screen = 1):
#arg = {left, right}
#a = sum(x) in current figure
#b = sum(x) in new figure
#c = sum(y) in current figure
#d = sum(y) in new figure
if screen == 2:
self.swap()
new_figure = figure()
new_figure.coords = self.curr.coords
a, b, c, d = 0, 0, 0, 0
#a, b, c, d = eval("new_figure." + arg + '()')
a, b, c, d = getattr(new_figure, arg)()
while a < b:
new_figure.move('up')
b -= 4
while a > b:
new_figure.move('down')
a -= 4
while c < d:
new_figure.move('left')
d -= 4
while c > d:
new_figure.move('right')
c -= 4
for i in range(len(new_figure.coords)):
x, y = new_figure.coords[i]
if x >= height or x < 0 or y >= wight or y < 0 or (self.screen[x][y] != ' ' and [x, y] not in self.curr.coords):
del new_figure
if screen == 2:
self.swap()
# print(1, end = '')
return False
for i in range(len(new_figure.coords)):
self.screen[self.curr.coords[i][0]][self.curr.coords[i][1]] = ' '
for i in range(len(new_figure.coords)):
self.screen[new_figure.coords[i][0]][new_figure.coords[i][1]] = ('#', self.curr_color)
del self.curr
self.curr = new_figure
if screen == 2:
self.swap()
return True
from IsarMain import IsarMain
from ParseObj import ParseObj
import figure
# from figure import figure1,figure2
# load rka file
import scipy.io as sio # MATLAB 5.0 MAT-file
rkafile = sio.loadmat("rkafile.mat")
rkaArr = rkafile['a']
# rkafile = 0 # this is a 10000x11 array
# IsarMain
XX, YY, Th1, Ph1, ISAR_VVdB, ISAR_VHdB, ISAR_HVdB, ISAR_HHdB = IsarMain(rkaArr)
# wavefront object read
obj = ParseObj("../kepce.obj")
# plot 1)graph and 2)object from top(do later)
rd: int = 40
figure.figure(obj, XX, YY, rd, Th1, Ph1, ISAR_VVdB, ISAR_VHdB, ISAR_HVdB,
ISAR_HHdB)
import os, sys, threading
import curses, colored
from time import clock, sleep, time
from figure import figure
monitor = curses.initscr()
wight = 9
height = 23
invisible = 4
figures = [figure() for i in range(7)]
figures[0].make([[0, 0], [0, 1], [0, 2], [0, 3]])
figures[1].make(([[0, 0], [0, 1], [0, 2], [1, 2]]))
figures[2].make(([[0, 0], [0, 1], [0, 2], [1, 1]]))
figures[3].make(([[0, 0], [0, 1], [1, 2], [1, 1]]))
figures[4].make(([[0, 2], [0, 1], [1, 0], [1, 1]]))
figures[5].make([[0, 0], [0, 1], [1, 1], [1, 0]])
figures[6].make(([[0, 0], [0, 1], [0, 2], [-1, 2]]))
log = open('log.txt' , 'w')
color = {}
color[figures[0]] = 'light_yellow'
color[figures[1]] = 'purpur'
color[figures[2]] = 'dark_blue'
color[figures[3]] = 'light_blue'
color[figures[4]] = 'green'
color[figures[5]] = 'red'
# Forward
L1 = np.dot(x, w1) + b1[0]
L1 = self.sigmoid(L1)
L2 = np.dot(L1, w2) + b2[0]
L2 = self.sigmoid(L2)
return np.argmax(L2, axis=1)
if __name__ == "__main__":
x_data = np.loadtxt('./data/exam_x.txt') # data
y_label = np.loadtxt('./data/exam_y.txt', dtype=int) # label
# x_data = np.loadtxt('./data/iris_x.txt') # data
# y_label = np.loadtxt('./data/iris_y.txt', dtype=int) # label
config = dict()
config['input_dim'] = 2
config['output_dim'] = 2
config['learning_rate'] = 0.01
config['nodes_num'] = 10
config['batch_size'] = 10
config['itrnum'] = 2000
NNmodel = NN(config)
loss = NNmodel.model(x_data, y_label)
# figure.lossfig(loss)
figure.figure(x_data, y_label, predict=NNmodel.predict, loss=loss)
#weight_conv2d_3, bias_conv2d_3 = model.get_layer('conv2d_3').get_weights()
#weight_conv2d_4, bias_conv2d_4 = model.get_layer('conv2d_4').get_weights()
#weight_conv2d_5, bias_conv2d_5 = model.get_layer('conv2d_5').get_weights()
plt.figure()
p1, = plt.plot(train_process.history['loss'])
p2, = plt.plot(train_process.history['val_loss'])
plt.legend([p1, p2], ['train_loss', 'val_loss'])
#val_process = model.evaluate(data_val, expect_val)
#predict_result = model.predict(data_test, batch_size=data_test.shape[0], verbose=0)
predict_result = model.predict(data_test,
batch_size=data_val.shape[0],
verbose=0)
from sklearn.metrics import mean_absolute_error
MAE_predict = mean_absolute_error(predict_result.flatten(),
expect_test.flatten())
figure(data_test, 50, 'Data test')
figure(predict_result, 50, 'Predict result')
figure(expect_test, 50, 'Expect test')
'''
####################显示每一层神经层的输出############################
layer_outputs = [layer.output for layer in model.layers[:5]]
activation_model = models.Model(inputs=model.input, outputs=layer_outputs)
activations = activation_model.predict(data_test)
##################################################################
'''
display('Project end. Good luck !!!')
data.trend_rule.update(season)
# ルール数を記録
CIM.act_rule_num.append(data.trend_rule.rule_num)
if cfg.SHOW_MODEL_DETAIL:
print("")
end = time.time()
print("合計時間：" + str(end - start))
rs = cfg.LSTM_REFERENCE_STEPS + 1
rt = cfg.REVEAL_TREND
# モデルの結果を出力
fg = figure("result/Research/research",
200,
cfg.SPAN,
data,
CIM,
reference_steps=rs,
reveal_trend=rt)
fg.savefig_result("PredictTrend")
fg.savefig_ruleweight("TrendRuleW")
fg.savefig_chosenrule("ChosenRule")
fg.savefig_compare_prediction("ComparePrediction")
fg.savefig_compare_prediction_ave("ComparePredictionAverage")
fg.savefig_rule_num("RuleMoving")
fg.save_config("config", cfg)
