def plot_nor_ms(songs):
songs_num=len(songs)
sum_play=np.array([0 for i in range(DAYS)])
sum_download=np.array([0 for i in range(DAYS)])
sum_collect=np.array([0 for i in range(DAYS)])
with open(SONG_P_D_C,'r') as fr:
songs_id=fr.readline().strip("\n")
while songs_id and songs_num>0:
play = list(map(int, fr.readline().strip("\n").split(",")))
download = list(map(int, fr.readline().strip("\n").split(",")))
collect = list(map(int, fr.readline().strip("\n").split(",")))
if songs_id in songs:
play=np.array(play)
sum_play+=play
songs_num-=1
songs_id=fr.readline().strip("\n")
p = plt.plot(sum_play, "bo", sum_play, "b-", marker="o")
#d = plt.plot(download, "ro", download, "r-", marker="o")
#c = plt.plot(collect, "go", collect, "g-", marker="o")
#plt.legend([p[1], d[1],c[1]], ["play", "download","collect"])
plt.lengend(p[1],["play"])
plt.title('SUM OF THE NORMAL MUSIC')
plt.xlabel('days')
plt.ylabel('times')
#plt.savefig(os.path.join(self.SONG_PLAY_FOLDER, songs_id+".png"))
plt.show()
def on_epoch_end(self, epoch, logs={}):
self.logs.append(logs)
self.x.append(self.i)
self.losses.append(logs.get("loss"))
self.val_losses.append(logs.get("val_loss"))
self.i = self.i + 1
clear_output(wait=True)
plt.plot(self.x, self.losses, label="loss")
plt.plot(self.x, self.var_losses, label="val_loss")
plt.lengend()
plt.show()
def plotRainDrops(dropInCircle, dropsOutOfCircle, lengthOfField=1, format = 'pdf'):
numberOfDropsInCircle = len(dropInCircle)
numberOfDropsOutCircle = len(dropInCircle)
numberOfDrops = numberOfDropsInCircle + numberOfDropsOutCircle
plt.figure()
plt.xlim(-lengthOfField/2, lengthOfField/2)
plt.ylim(-lengthOfField/2, lengthOfField/2)
plt.scatter([e[0] for e in dropInCircle], [e[1] for e in dropInCircle], color = 'black', label = 'Drops en circle')
plt.scatter([e[0] for e in dropsOutOfCircle], [e[1] for e in dropsOutOfCircle], color = 'red', label = 'Drops fuera circle')
plt.lengend(loc='center')
plt.title("%s drop: %s landed in circle, estimating $\pi$ as %.4f" %(numberOfDrops, numberOfDropsInCircle, 4 * numberOfDropsInCircle/numberOfDrops))
plt.savefig("%s_drops.%s" % (numberOfDrops, format))
def plot_acc(history):
print("plot starts")
acc = history.history['acc']
val_acc = history.history['val_acc']
epochs = range(1, len(acc) + 1)
# "bo" is for "blue dot"
plt.plot(epochs, acc, 'bo', label="Training acc")
# b is for "solid blue line"
plt.plot(epochs, val_acc, "b", label="Validation acc")
plt.title('Training and validation acc')
plt.xlabel('Epochs')
plt.ylabel('ACC')
plt.lengend()
plt.show()
print("plot finished!")
epoch_set = []
init = tf.initializer_all_variables()
with tf.Session() as sess:
sess.run(init)
for epoch in range(training_epochs):
avg_cost = 0.0
total_batch = int(mnist.train.num_examples / batch_size)
for i in range(total_batch):
batch_xs, batch_ys = mnist.train.next_batch(batch_size)
sess.run(optimizer, feed_dict={x: batch_xs, y: batch_ys})
avg_cost += sess.run(cost, feed_dict={x: batch_xs, y: batch_ys}) / total_batch
if epoch % display_step == 0:
print "Epoch: ", '%04d' %(epoch+1), "cost=", '{:.9f}'.format(avg_cost)
avg_set.append(avg_cost)
epoch_set.append(eopch+1)
print ('Training phase finished')
plt.plot(epoch_set, avg_set, 'o', label='MLP Training Phase')
plt.ylabel('cost')
plt.xlabel('epoch')
plt.lengend()
plt.show()
#Test model
correct_prediction = tf.equal(tf.argmax(output_layer, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float))
print('Model Accuracy', accuracy.eval({x: mnist.test.images, y: mnist.test.labels}))
loss.backward()
# 参数更新
optimizer.step()
import matplotlib.pyplot as plt
# 绘制不同迭代过程的loss
x = np.arange(max_epoch)
y = np.array(iter_loss)
plt.plot(x,y)
plt.title('Loss Value in all interations')
plt.xlabel('Interation')
plt.ylabel('Mean loss Value')
plt.show
# 测试
output = model(text_x)
predict_list = output.detach().numpy()
print(predict_list)
# 真实值与预测值的散点图
x = np.arange(text_x.shape[0])
y1 = np.arange(predict_list)
y2 = np.arange(text_y)
line1 = plt.scatter(x,y1,c='red',Label='predict')
line2 = plt.scatter(x,y2,c='yellow',Label='real')
plt.lengend(loc = 'best')
plt.title('Prediction Vs Real')
plt.ylabel('House Price')
plt.show()
# plt.hist(train_transformed[:, i], alpha=0.3, label="Latent User " + str(i + 1), range=(0, 1), bins=20)
# plt.xlabel("User Proportion from Latent User i", fontsize=20)
# plt.ylabel("Count", fontsize=20)
# plt.tick_params(labelsize=15)
# plt.legend()
# plt.show(block=False)
plt.figure()
targets = [1, 0]
colors = ['r', 'b']
for target, color in zip(targets, colors):
indicesToKeep = train['sex_m'] == target
plt.scatter(train_transformed[indicesToKeep, 0],
train_transformed[indicesToKeep, 1],
c=color)
plt.lengend(targets)
plt.xlabel('component 1')
plt.ylabel('component 2')
plt.show(block=False)
if n['name'] == 'PCA':
kmeans = KMeans().fit(train_transformed) # n_clusters=2
# cluster_transformed = kmeans.transform()
centroids = kmeans.cluster_centers_
print("Train - Average Log-Likelihood with Kmeans: ",
kmeans.score(train_transformed) / train.shape[0])
print("Test - Average Log-Likelihood with Kmeans: ",
kmeans.score(test_transformed) / test.shape[0])
plt.figure()
plt.plot(train_transformed[:, 0],
train_transformed[:, 1],
def main():
global args, start_epoch, best_acc1
args = config()
if args.cuda and not torch.cuda.is_available():
raise Exception('No GPU found, please run without --cuda')
print('\n=> Build ResNet..')
model = mo.ResNet50()
print(model)
print('==> Complete build')
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(),
lr=args.lr,
momentum=args.momentum,
weight_decay=args.weight_decay,
nesterov=True)
start_epoch = 0
n_retrain = 0
if args.cuda:
torch.cuda.set_device(args.gpuids[0])
with torch.cuda.device(args.gpuids[0]):
model = model.cuda()
criterion = criterion.cuda()
model = nn.DataParallel(model,
device_ids=args.gpuids,
output_device=args.gpuids[0])
cudnn.benchmark = True
# checkpoint file
ckpt_dir = pathlib.Path('checkpoint')
ckpt_file = ckpt_dir / args.dataset / args.ckpt
# for resuming training
if args.resume:
if isfile(ckpt_file):
print('\n==> Loading Checkpoint \'{}\''.format(args.ckpt))
checkpoint = load_model(model, ckpt_file, args)
start_epoch = checkpoint['epoch']
optimizer.load_state_dict(checkpoint['optimizer'])
print('==> Loaded Checkpoint \'{}\' (epoch {})'.format(
args.ckpt, start_epoch))
else:
print('==> no checkpoint found \'{}\''.format(args.ckpt))
return
# Data loading
print('\n==> Load data..')
train_loader, val_loader = DataLoader(args.batch_size, args.workers,
args.datapath, args.cuda)
# for evaluation
if args.evaluate:
if isfile(ckpt_file):
print('\n==> Loading Checkpoint \'{}\''.format(args.ckpt))
checkpoint = load_model(model, ckpt_file, args)
print('==> Loaded Checkpoint \'{}\' (epoch {})'.format(
args.ckpt, start_epoch))
# evaluate on validation set
print('\n===> [ Evaluation ]')
start_time = time.time()
acc1, acc5 = validate(val_loader, model, criterion)
elapsed_time = time.time() - start_time
print('====> {:.2f} seconds to evaluate this model\n'.format(
elapsed_time))
return
else:
print('==> no checkpoint found \'{}\''.format(args.ckpt))
return
# train...
train_time = 0.0
validate_time = 0.0
lr = args.lr
list_Acc1 = []
list_Acc5 = []
list_epoch = []
for epoch in range(start_epoch, args.epochs):
adjust_learning_rate(optimizer, epoch, lr)
print('\n==> Epoch: {}, lr = {}'.format(
epoch, optimizer.param_groups[0]["lr"]))
# train for one epoch
print('===> [ Training ]')
start_time = time.time()
acc1_train, acc5_train = train(train_loader,
epoch=epoch,
model=model,
criterion=criterion,
optimizer=optimizer)
elapsed_time = time.time() - start_time
train_time += elapsed_time
print(
'====> {:.2f} seconds to train this epoch\n'.format(elapsed_time))
# evaluate on validation set
print('===> [ Validation ]')
start_time = time.time()
acc1_valid, acc5_valid = validate(val_loader, model, criterion)
elapsed_time = time.time() - start_time
validate_time += elapsed_time
print('====> {:.2f} seconds to validate this epoch\n'.format(
elapsed_time))
# remember best Acc@1 and save checkpoint
is_best = acc1_valid > best_acc1
best_acc1 = max(acc1_valid, best_acc1)
state = {
'epoch': epoch + 1,
'model': model.state_dict(),
'optimizer': optimizer.state_dict()
}
save_model(state, epoch, is_best, args)
list_Acc1.append(acc1_valid)
list_Acc5.append(acc5_valid)
list_epoch.append(epoch)
plt.plot(list_epoch, list_Acc1)
plt.plot(list_epoch, list_Acc5)
plt.lengend(['ACC1', 'ACC5'])
avg_train_time = train_time / (args.epochs - start_epoch)
avg_valid_time = validate_time / (args.epochs - start_epoch)
total_train_time = train_time + validate_time
print('====> average training time per epoch: {:,}m {:.2f}s'.format(
int(avg_train_time // 60), avg_train_time % 60))
print('====> average validation time per epoch: {:,}m {:.2f}s'.format(
int(avg_valid_time // 60), avg_valid_time % 60))
print('====> training time: {}h {}m {:.2f}s'.format(
int(train_time // 3600), int((train_time % 3600) // 60),
train_time % 60))
print('====> validation time: {}h {}m {:.2f}s'.format(
int(validate_time // 3600), int((validate_time % 3600) // 60),
validate_time % 60))
print('====> total training time: {}h {}m {:.2f}s'.format(
int(total_train_time // 3600), int((total_train_time % 3600) // 60),
total_train_time % 60))
df[abbv] = (df[abbv] - df[abbv][0] / df[abbv][0] * 100)
if main_df.empty:
main_df = df
else:
main_df = main_df.join(df)
print(main_df.head())
# Printing out Pickle
pickle_out = open('fiddy_states.pickle', 'wb')
pickle.dump(main_df, pickle_out)
pickle_out.close()
def HPI_Benchmark():
df = quandl.get("FMAC/HPI_USA", authtoken=api_key)
df["United States"] = (
df["United States"] -
df["United States"][0] / df["United States"][0] * 100)
return df
fig = plt.figure()
ax1 = plt.subplot2grid((1, 1), (0, 0))
HPI_data = pd.read_pickle('pickle.pickle')
HPI_data.plot()
plt.lengend().remove()
plt.show()
import matplotlib.pyplot as plt
import pickle
import pandas as pd
#grab_initial_state_data()
fig = plt.figure()
ax1 = plt.subplot2grid((1,1), (0,0))
HPI_data = pd.read_pickle('fiddy_states.pickle')
HPI_data['TX1yr'] = HPI_data['TX'].resample('A', how='mean')
print(HPI_data[['TX','TX1yr']].head())
HPI_data[['TX','TX1yr']].plot(ax=ax1)
plt.lengend(loc=4)
plt.show()
