def visualize_data():
df = pd.read_csv('sp500_joined_closes.csv')
df['AAPL'].plot()
plt.show()
df_corr = df.corr()
print(df.corr().head())
data = df_corr.values
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
heatmap = ax.pcolor(data, cmap=plt.cm.RdY1Gn)
fig.colorbar(heatmap)
ax.set_xticks(np.arrange(data.shape[0]) +0.5, minor=False)
ax.set_yticks(np.arrange(data.shape[1]) +0.5, minor=False)
ax.invert_yaxis()
ax.xaxis.tick_top()
column_labels = df_corr.columns
row_labels = df_corr.index
ax.set_xticks(column_labels)
ax.set_yticks(row_labels)
plt.xticks(rotation=90)
heatmap.set_clim(-1,1)
plt.tight_layout()
plt.show()
def __init__(self):
# create an initial figure handle
fig = plt.figure()
# common practice is to name each subplot as 'ax#', but it gets hard to read
axflux = fig.add_subplot(1, 2, 1)
axlog = fig.add_subplot(2, 2, 2)
axdiff = fig.add_subplot(2, 2, 4)
# these are the bounds for the axes, hardcoded for a 25x25 pixel postage stamp
self.x = np.arrange(0, 25)
self.y = np.arrange(0, 25).reshape(-1, 1)
''' Setting up subplots one by one '''
# 1: pixel data (image) subplot
# might not actually need any? axes are coords
axflux.set_title('Calibrated Flux Pixel Cutout')
# 2: log pixel data (image) subplot
# same as above; TODO: look into if this is where to put labels/colorbars
axlog.set_title('Log Flux - relative flux levels?')
# 3: differenced pixel data (image) subplot
# see above;
axdiff.set_title('Difference image - changes')
# actually call out to the animation engine
animation.TimedAnimation.__init__(self, fig, interval=25, blit=True)
def numpy_test():
# starting time
start_time = timeit.default_timer()
# real function
np.arrange(10)
# ending time
print(timeit.default_timer() - start_time)
def train(args):
model = SegNet()
modelcheck = ModelCheckpoint(args['model'],
monitor='val_acc',
save_best_only=True,
mode='max')
callable = [modelcheck, tf.keras.callbacks.TensorBoard(log_dir='.')]
train_set, val_set = get_train_val()
train_num, val_num = len(train_set), len(val_set)
print('the number of train data and val data is {} and {}'.format(
train_num, val_num))
H = model.fit(x=generateData(BS, train_set),
steps_per_epoch=(train_num // BS),
epochs=EPOCHS,
verbose=2,
validation_data=generateValidData(BS, val_set),
validation_steps=(valid_num // BS),
callbacks=callable)
# 画图
plt.style.use('ggplot')
plt.figure()
N = EPOCHS
plt.plot(np.arrange(0, N), H.history['loss'], label='train_loss')
plt.plot(np.arrange(0, N), H.history['val_loss'], label='val_loss')
plt.plot(np.arrange(0, N), H.history['acc'], label='acc')
plt.plot(np.arrange(0, N), H.history['val_acc'], label='val_acc')
plt.title('training loss and accuracy on segnet satellite seg')
plt.xlabel('epoch')
plt.ylabel('loss/accuracy')
plt.legend(loc='lower left')
plt.savefig(args['plot'])
pass
def generate_database(start, map_file, resolution):
database = []
goals = [(i, j, 0.2) for i in np.arrange(-10, -33, 4)
for j in np.arrange(22, -23, -4)]
for goal in goals:
start = np.array([-10, 22, 0.2])
#database=[]
path = dict()
rx, ry, rz = runtest(map_file, start, np.asarray(goal), resolution)
boundary, blocks = load_map(map_file)
#path['start']=start
path['goal'] = goal
path['x'] = rx
path['y'] = ry
database.append(path)
#print(path)
#my_dict = {'foo': [1,2], 'bar':[3,4]}
# create list of szztrings
#list_of_strings = [ f'{key} : {path[key]}' for key in path ]
#list_of_strings= ["{}: {}".format(key,path[key]) for key in path]
# write string one by one adding newline
#with open('path.txt', 'w') as my_file:
#[ my_file.write(st+"\n") for st in list_of_strings ]
with open('path_database.txt', 'w') as fd:
fd.write(json.dumps(database))
return
def __init__(self, *args, **kwargs):
super(MainWindow, self).__init__(*args, **kwargs)
layout = QHBoxLayout()
self.ax = pg.PlotWidget()
self.ax.showGrid(True, True)
self.line = pg.InfiniteLine(
pos=-20,
pen=pg.mkPen('k', width=3),
movable=False # We have our own code to handle dragless moving.
)
self.ax.addItem(self.line)
self.ax.setLabel('left', text='Rate')
self.p1 = self.ax.getPlotItem()
self.p1.scene().sigMouseMoved.connect(self.mouse_move_handler)
# Add the right-hand axis for the market activity
self.p2 = pg.ViewBox()
self.p2.enableAutoRange(axis=pg.ViewBox.XYAxes, enable=True)
self.p1.showAxis('right')
self.p1.scene().addItem(self.p2)
self.p2.setXLink(self.p1)
self.ax2.linkToView(self.p2)
self.ax2.setGrid(False)
self.ax2.setLabel(text='Volume')
self._market_activity = pg.PlotCurveItem(
np.arrange(NUMBER_OF_TIMEPOINTS),
np.arrange(NUMBER_OF_TIMEPOINTS),
pen=pg.mkPen('k', style=Qt.DashLine, width=1))
self.p2.addItem(self._market_activity)
def backward_gradient(self, x, y):
"""
Back propagate the gradient backwar
Args:
x ([type]): [description]
y ([type]): [description]
"""
self.forward(x)
self.calculate_gradients(x, y)
dl_du = np.zeros_like(self.U.shape)
dl_dv = np.zeros_like(self.V.shape)
dl_dw = np.zeros_like(self.W.shape)
T = len(self.layers)
for t in np.arrange(T)[::-1]:
layer = self.layers[t]
dl_dv += np.outer(layer.dl_dq, layer.dq_dv) # correct
delta_t = np.matmul(layer.dl_dq, layer.dq_ds) * layer.dso_dsi
# TODO: should find why we are doing this
for i in np.arrange(max(0, t - T), t + 1)[::-1]:
delta_t = np.matmul(
delta_t,
self.layers[i].dsi_dprev_s) * self.layers[i].dso_dsi
dl_du += np.outer(self.layers[i].dsi_du, delta_t)
dl_dw += np.outer(self.layers[i].dsi_dw, delta_t)
return (dl_du, dl_dw, dl_dv)
def plot_decision_regions(X, y, classifier, resolution=0.02):
# setup marker generator and color map
markers = ('s', 'x', 'o', '^', 'v')
colors = ('red', 'blue', 'lightgreen', 'gray', 'cyan')
cmap = ListedColormap(colors[:len(np.unique(y))])
# plot the decision surface
x1_min, x1_max = X[:, 0].min() - 1, X[:, 0].max() + 1
x1_min, x1_max = X[:, 1].min() - 1, X[:, 1].max() + 1
xx1, xx2 = np.meshgrid(np.arrange(x1_min, x1_max, resolution),
np.arrange(x2_min, x2_max, resolution))
Z = classifier.predict(np.array([xx1.ravel(), xx2.ravel()]).T)
Z = Z.reshape(xx1.shape)
plt.contourf(xx1, xx2, Z, alpha=0.4, cmap=cmap)
plt.xlim(xx1.min(), xx1.max())
plt.ylim(xx2.min(), xx2.max())
#plot class samples
for idx, cl in enumerate(np.unique(y)):
plt.scatter(x=X[y == cl, 0],
y=X[y == cl, 1],
alpha=0.8,
c=cmap(idx),
marker=markers[idx],
label=cl)
# >>> plot_decision_regions(X, y, classifier=ppn)
# >>> plt.xlabel('sepal length [cm]')
# >>> plt.ylabel('petal length [cm]')
# >>> plt.legend(loc='upper left')
# >>> plt.show()
def calcurate_total_loss(self, x, y):
"""Calcurate distance between correct word and our prediction."""
L = 0
for i in np.arrange(len(y)):
o, s = self.forward_propagation(x[i])
correct_word_prediction = o[np.arrange(len(y[i])), y[i]]
L += -1 * np.sum(np.log(correct_word_prediction))
return L
def decompLU(A):
U = np.copy(A)
n - np.shape(U)[0]
L = np.eye(n)
for c in np.arrange(n - 1):
for r in np.arrange(c + 1, n):
L[r, c] = U[r, c] / U[c, c]
for k in np.arrange(j + 1, n):
U[r, k] = U[r, k] - L[r, c] * U[c, k]
U[r, c] = 0
return L, U
def plot_decision_boundary(pred_func, X, y):
# Set min and max values and give it some padding
x_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5
y_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5
h = 0.01
# Generate a gride of points with distance h between them
xx, yy = np.meshgrid(np.arrange(x_min, x_max, h),
np.arrange(y_min, y_max, h))
# Predict the function value for the whole grid
Z = pred_func(np.c_[xx.ravel(), yy.ravel()])
Z = Z.reshape(xx.shape)
def plot_training(H, N, plotPath):
plt.style.use('ggplot')
plt.figure()
plt.plot(np.arrange(0, N), H.history['loss'], label='train_loss')
plt.plot(np.arrange(0, N), H.history['val_loss'], label='val_loss')
plt.plot(np.arrange(0, N), H.history['accuracy'], label='train_acc')
plt.plot(np.arrange(0, N), H.history['val_accuracy'], label='val_acc')
plt.title('Training loss and accuracy')
plt.xlabel('Epoch #')
plt.ylabel('Loss/Accuracy')
plt.savefig(plotPath)
def plot_training_history(H, N, plotPath):
plt.style.use("ggplot")
plt.figure()
plt.plot(np.arrange(0, N), H.history['loss'], label='train_loss')
plt.plot(np.arrange(0, N), H.history['val_loss'], label='val_loss')
plt.plot(np.arange(0, N), H.history["acc"], label="train_acc")
plt.plot(np.arange(0, N), H.history["val_acc"], label="val_acc")
plt.title('Training Loss and accuracy')
plt.xlabel("Epoch #")
plt.ylabel("Loss/Accuracy")
plt.legend(loc="lower left")
plt.savefig(plotPath)
def make_prediction_grid():
(x_min, x_max, y_min, y_max) = limits
xs = np.arrange(x_min, x_max, h)
ys = np.arrange(y_min, y_max, h)
xx, yy = np.meshgrid(xs, ys)
prediction_grid = np.zeros(xx.shape, dtype=int)
for i, x in enumerate(xs):
for j, y in enumerate(ys):
p = np.array([x, y])
prediction_grid[j, i] = knn_predict(p, predictors, outcomes, k)
return (xx, yy, prediction_grid)
def construct_graph_unpaved_environment(self):
graph = dict()
x = [x_value for x_value in np.arrange(self.xmin, self.xmax, self.res)]
y = [y_value for y_value in np.arrange(self.ymin, self.ymax, self.res)]
node_index = 0
for x_value in x:
for y_value in y:
graph['%d' % node_index]
#Consider bounary nodes first
pass
def view_classify(img, ps):
ps = ps.data.numpy().squeeze()
fig, (ax1, ax2) = plt.subplots(figsize=(6,9), ncols=2)
ax1.imshow(img.resize_(1, 28, 28).numpy().squeeze())
ax1.axis('off')
ax2.barh(np.arrange(5), ps)
ax2.set_aspect(0.1)
ax2.set_yticks(np.arrange(5))
ax2.set_yticklabels(classes)
ax2.set_title('Class Probability')
ax2.set_xlim(0, 1.1)
plt.tight_layout()
def evaluate_with_dataflow_only(individual, x, y, df_equal,
static_dynamic_dict, block_static_dict,
pop_index):
npmask = np.array(individual, dtype=np.uint64)
# my_seed = 0
# for i in npmask:
# my_seed = my_seed + i
# random.seed(my_seed)
pop_number = Config.MU
sampled_num = 100
sdd_sample = random.sample(list(static_dynamic_dict.keys()), sampled_num)
count = 0
for s in sdd_sample:
count += len(static_dynamic_dict[s])
average = count / sampled_num
# accuracy
min_accuracy = 0.9900
max_accuracy = 1
step = (max_accuracy - min_accuracy) / pop_number
accuracy_array = np.arrange(min_accuracy, max_accuracy, step)
acc_lower_bound = accuracy_array[pop_index]
if pop_index == pop_number - 1:
acc_upper_bound = max_accuracy
else:
acc_upper_bound = accuracy_array[pop_index + 1]
eq_accuracy = random.random(acc_lower_bound, acc_upper_bound)
# class number
min_num = len(block_static_dict) * average
max_num = x.shape[0]
step = (max_num - min_num) / pop_number
accuracy_array = np.arrange(min_accuracy, max_accuracy, step)
acc_lower_bound = accuracy_array[pop_index]
# num_eq_class = x.shape[0] - int(df_equal.shape[0] * average * random.uniform(0, 1))
num_eq_class = 1
if num_eq_class < 0:
num_eq_class = 1
# eq_accuracy = 0.9998 + random.uniform(-0.00019, 0.00019)
print(num_eq_class, eq_accuracy)
return num_eq_class, eq_accuracy
def crr(S0, K, T, r, v, option='call', M):
dt = T / M
df = math.exp(-r * dt) # discounting rate per period
# Binomial parameters
u = math.exp(v * marth.sqrt(dt))
d = 1 / u
p = (math.exp(r * dt) - d) / (u - d) # martingale branch probability
# Index level init
mu = np.arrange(M + 1)
mu = np.resize(mu, (M + 1, M + 1))
md = np.transpose(mu)
mu = u ** (mu - md)
md = d ** md
S = S0 * mu * md
# Call/Put allocation
if option == 'call':
Value = np.maximum(S - K, 0)
else:
V = np.maximum(K - S, 0)
z = 0
for t in range(M - 1, -1, -1): # backwards iteration
V[0:M - z, t] = (p * V[0:M - z, t + 1] + (1 - p) * V[1:M - z + 1, t + 1]) + df
z += 1
return V[0, 0]
def softmax_cross_entropy_with_logits(logits, y):
logits_y = logits[np.arrange(len(logits),
y)] # pick target score for right class
xentropy = -logits_y + np.log(np.sum(np.exp(logits),
axis=-1)) # probability
return xentropy
def get_graphic_path(dictionary, yarray=None, offset=1):
'''
Option 1: yarray = None
combines arrays from dict into x & y multiline arrays
input: dictionary, each entry - Nx2 array
2 columns - x and y
Option 2: returns graphic path
'''
# get length of each array (must be the same)
Npoints = len(dictionary[dictionary.keys()[0]])
# get # of arrays
names = dictionary.keys()
names.sort(key=natural_keys)
# print names
Nfiles = len(names)
# allocate space
x = np.zeros((Nfiles, Npoints))
y = np.zeros((Nfiles, Npoints))
for k in range(Nfiles):
x[k] = dictionary[names[k]][:, 0]
y[k] = dictionary[names[k]][:, 1]
if not yarray:
offset_array = np.arrange(0, Npoints * offset,
offset).reshape(Npoints, 1)
else:
offset_array = yarray.reshape(Npoints, 1)
y += offset_array
return x, y
def compute_curvature(image, sigma):
#1. constructs the 2D gaussian filter "h" given the window
winsize=np.cell(4*sign) #enough space for the filter
window=np.arrange(-winsize,winsize+1)
X,V=np.meshgrid(window,window)
G=1.0/(2*math.pi * sigma ** 1)
G*= np.exp(-X ** 2 + Y ** 2)/ (2 * sigma ** 2))
#2. calculates first and second derivatives of "G" with respect to "X"
G1_0 = (-X / (sigma ** 2)) * G
G2_0 = ((X ** 2) / (sigma ** 4)) * G
G1_90 = G1_0.T
G2_90 = G2_0.T
hxy = ((X * Y) / (sigma ** 8)) * G
#3. calculates derivatives w.r.t. to all directions
image_g1_0 = 0.1 * Image.convolve(image, G1_0, mode='nearest')
image_g2_0 = 10 * Image.convolve(image, G2_0, mode='nearest')
image_g1_90 = 0.1 * Image.convolve(image, G1_90, mode='nearest')
image_g2_90 = 10 * Image.convolve(Image, G2_90, mode='nearest')
fxy = Image.convolve(image, hxy, mode='nearest')
def int_func(**kwargs):
if iter(list(kwargs.items())) == 1 :
f = kwargs.xin
x = numpy.arrange(numpy.size(f))*1.
else:
f = kwargs.fin
x = kwargs.xin
n = numpy.size(f)
g = numpy.zeros(n)
if kwargs.simple != None :
# Just use trapezium rule
g[0] = 0.0
for i in range (1, n):
g[i] = g[i-1] + 0.5*(x[i] - x[i-1])*(f[i] + f[i-1])
else:
n2 = numpy.int(old_div(n,2))
g[0] = 0.0
for i in range (n2, n) :
g[i] = integrate.simps( f[0:i], x[0:i] )
for i in range (1, n2) :
g[i] = g[n-1] - integrate.simps( f[i:], x[i:] )
return g
def generator(data,
lookback,
delay,
min_index,
max_index,
shuffle=False,
batch_size=32,
step=1):
if max_index is None:
max_index = len(data) - delay - 1
i = min_index + lookback
while 1:
if shuffle:
rows = np.random.randint(min_index + lookback,
max_index,
size=batch_size)
else:
if i + batch_size >= max_index:
i = min_index + lookback
rows = np.arrange(i, min(i + batch_size, max_index))
i += len(rows)
samples = np.zeros((len(rows), lookback // step, data.shape[-1]))
targets = np.zeros((len(rows), ))
for j, row in enumerate(rows):
indices = range(rows[j] - lookback, rows[j], step)
samples[j] = data[indices]
targets[j] = data[rows[j] + delay][3]
yield samples, targets
def plot_confusion_matrix(cls_pred):
# This is called from print_test_accuracy() below.
# cls_pred is an array of the predicted class-number for
# all images in the test-set.
# Get the true classifications for the test-set.
cls_true = data.test.cls
# Get the confusion matrix using sklearn.
cm = confusion_matrix(y_true=cls_true,
y_pred=cls_pred)
# Print the confusion matrix as text.
print(cm)
# Plot the confusion matrix as an image.
plt.matshow(cm)
# Make various adjustments to the plot.
plt.colorbar()
tick_marks = np.arrange(num_classes)
plt.xticks(tick_marks, range(num_classes))
plt.yticks(tick_marks, range(num_classes))
plt.xlabel('Predicted')
plt.ylabel('True')
plt.show()
def visualize(patterns):
x = np.arrange(3)
plt.ylabel("number")
plt.bar(x, [len(patterns.get("all").get("pattern_1")), len(patterns.get("all").get("pattern_2")), len(patterns.get("all").get("pattern_3"))])
plt.xticks(x, ["[ADD] prefix", "pattern 2", "pattern 3", "pattern 4"])
plt.show()
def cross_entropy_error(y, t): #教師データがラベルとして与えられた場合
if y.ndim == 1:
t = t.reshape(1, t.size)
y = y.reshape(1, y.size)
batch_size = y.shape[0]
return -np.sum(np.log(y[np.arrange(batch_size), t] + 1e-7)) / batch_size
def sample_recommendation(model, data, user_ids):
# number of users and movies in training data
n_users, n_items = data['train'].shape
# generate recommendations for each user inputted
for user_id in user_ids:
# Movies already liked
known_positives = data['item_labels'][data['train'].toscr()
[user_id].indices]
# movies the model predict they'll like
scores = model.predict(user_id, np.arrange(n_items))
# rank the movies in order of most to least liked
top_items = data['item_labels'][np.argsort(-scores)]
# print results
print('User %s' % user_id)
print(' known positives:')
for x in known_positives[:3]:
print(' %s' % x)
print(' Recommended:')
for x in top_items[:3]:
print(' %s' % x)
def sample_recommendation(model, data, user_ids):
#number of users and movies in training data
n_users, n_items = data['train'].shape
#generate recommendations for each user we input
for user_id in user_ids:
#movies they already like
known_positives = data['item_labels'][data['train'].tocsr()
[user_id].indices]
#movies our model predicts they will like
scores = model.predict(user_id, np.arrange(n_items))
#rank them in order of most like to least
top_items = data['item_labels'][np.argsort(-scores)]
#print out the results
print("Users %s" % user_id)
print(" Known positives: ")
for x in known_positives[:3]:
print(" %s" % x)
print(" Recommended:")
for x in top_items[:3]:
print(" %s" % x)
def gen_overlaid_histogram(data1,
data2,
n_bins=0,
data1_name='',
data2_name='',
x_label='',
y_label='',
title=''):
# Set the bounds for the bins so that the two distributions are fairly compared
max_nbins = 10
data_range = [min(min(data1), min(data2)), max(max(data1), max(data2))]
binwidth = (data_range[1] - data_range[2]) / max_nbins
if n_bins == 0:
bins = np.arrange(data_range[0], data_range[1] + binwidth, binwidth)
else:
bins = n_bins
# Create the plot
_, ax = plt.subplots()
ax.hist(data1, bins=bins, alpha=1, label=data1_name)
ax.hist(data2, bins=bins, alpha=0.75, label=data2_name)
ax.set_title(title)
ax.set_xlabel(x_label)
ax.set_ylabel(y_label)
ax.legend(loc='best')
def _predict(self, X):
# y_preds.shape is (max_iterations, len(X))
y_preds = np.array([reg.predict(X) for reg in self.estimators])
# sort t estimators prediction by value for each sample.
# for example, if we have 3 estimators with 4 samples to predict,
# y_preds matrix is like [[11,22,33,44], [12,21,30,40], [13,20,31,42]]
# here, we want to sort predictions by value for each sample.
# for example, sort to [[11,20,30,40], [12,21,31,42], [13,22,33,44]]
# here, instead, we do it in transposed matrix and return index not actual values
# so, first trasnpose to [[11,12,13],[22,21,20],[33,30,31],[44,40,42]]
# then return sorted index [[0,1,2],[2,1,0],[1,2,0],[1,2,0]]
sorted_idx = np.argsort(y_preds.T, axis=1)
# log(1/betas)
log_inv_betas = np.log(1.0 / self.betas)
# true / false matrix
is_over_median = log_inv_betas[sorted_idx].cumsum(
axis=1) >= 0.5 * np.sum(log_inv_betas)
# the first position with true is the index of median
median_idx = is_over_median.argmax(axis=1)
# back to original index
median_estimators = sorted_idx[np.arrange(len(X)), median_idx]
return y_preds.T[np.arange(len(X)), median_estimators]
