In [0]:
#@title Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
In [0]:
import numpy as np
import matplotlib.pyplot as plt
import tensorflow as tf
keras = tf.keras
In [0]:
def plot_series(time, series, format="-", start=0, end=None, label=None):
plt.plot(time[start:end], series[start:end], format, label=label)
plt.xlabel("Time")
plt.ylabel("Value")
if label:
plt.legend(fontsize=14)
plt.grid(True)
def trend(time, slope=0):
return slope * time
def seasonal_pattern(season_time):
"""Just an arbitrary pattern, you can change it if you wish"""
return np.where(season_time < 0.4,
np.cos(season_time * 2 * np.pi),
1 / np.exp(3 * season_time))
def seasonality(time, period, amplitude=1, phase=0):
"""Repeats the same pattern at each period"""
season_time = ((time + phase) % period) / period
return amplitude * seasonal_pattern(season_time)
def white_noise(time, noise_level=1, seed=None):
rnd = np.random.RandomState(seed)
return rnd.randn(len(time)) * noise_level
def window_dataset(series, window_size, batch_size=32,
shuffle_buffer=1000):
dataset = tf.data.Dataset.from_tensor_slices(series)
dataset = dataset.window(window_size + 1, shift=1, drop_remainder=True)
dataset = dataset.flat_map(lambda window: window.batch(window_size + 1))
dataset = dataset.shuffle(shuffle_buffer)
dataset = dataset.map(lambda window: (window[:-1], window[-1]))
dataset = dataset.batch(batch_size).prefetch(1)
return dataset
def model_forecast(model, series, window_size):
ds = tf.data.Dataset.from_tensor_slices(series)
ds = ds.window(window_size, shift=1, drop_remainder=True)
ds = ds.flat_map(lambda w: w.batch(window_size))
ds = ds.batch(32).prefetch(1)
forecast = model.predict(ds)
return forecast
In [0]:
time = np.arange(4 * 365 + 1)
slope = 0.05
baseline = 10
amplitude = 40
series = baseline + trend(time, slope) + seasonality(time, period=365, amplitude=amplitude)
noise_level = 5
noise = white_noise(time, noise_level, seed=42)
series += noise
plt.figure(figsize=(10, 6))
plot_series(time, series)
plt.show()
In [0]:
split_time = 1000
time_train = time[:split_time]
x_train = series[:split_time]
time_valid = time[split_time:]
x_valid = series[split_time:]
In [0]:
keras.backend.clear_session()
tf.random.set_seed(42)
np.random.seed(42)
window_size = 30
train_set = window_dataset(x_train, window_size, batch_size=128)
model = keras.models.Sequential([
keras.layers.Lambda(lambda x: tf.expand_dims(x, axis=-1),
input_shape=[None]),
keras.layers.SimpleRNN(100, return_sequences=True),
keras.layers.SimpleRNN(100),
keras.layers.Dense(1),
keras.layers.Lambda(lambda x: x * 200.0)
])
lr_schedule = keras.callbacks.LearningRateScheduler(
lambda epoch: 1e-7 * 10**(epoch / 20))
optimizer = keras.optimizers.SGD(lr=1e-7, momentum=0.9)
model.compile(loss=keras.losses.Huber(),
optimizer=optimizer,
metrics=["mae"])
history = model.fit(train_set, epochs=100, callbacks=[lr_schedule])
In [0]:
plt.semilogx(history.history["lr"], history.history["loss"])
plt.axis([1e-7, 1e-4, 0, 30])
In [0]:
keras.backend.clear_session()
tf.random.set_seed(42)
np.random.seed(42)
window_size = 30
train_set = window_dataset(x_train, window_size, batch_size=128)
valid_set = window_dataset(x_valid, window_size, batch_size=128)
model = keras.models.Sequential([
keras.layers.Lambda(lambda x: tf.expand_dims(x, axis=-1),
input_shape=[None]),
keras.layers.SimpleRNN(100, return_sequences=True),
keras.layers.SimpleRNN(100),
keras.layers.Dense(1),
keras.layers.Lambda(lambda x: x * 200.0)
])
optimizer = keras.optimizers.SGD(lr=1.5e-6, momentum=0.9)
model.compile(loss=keras.losses.Huber(),
optimizer=optimizer,
metrics=["mae"])
early_stopping = keras.callbacks.EarlyStopping(patience=50)
model_checkpoint = keras.callbacks.ModelCheckpoint(
"my_checkpoint", save_best_only=True)
model.fit(train_set, epochs=500,
validation_data=valid_set,
callbacks=[early_stopping, model_checkpoint])
In [0]:
model = keras.models.load_model("my_checkpoint")
In [0]:
rnn_forecast = model_forecast(
model,
series[split_time - window_size:-1],
window_size)[:, 0]
In [0]:
plt.figure(figsize=(10, 6))
plot_series(time_valid, x_valid)
plot_series(time_valid, rnn_forecast)
In [0]:
keras.metrics.mean_absolute_error(x_valid, rnn_forecast).numpy()
In [0]:
def seq2seq_window_dataset(series, window_size, batch_size=32,
shuffle_buffer=1000):
series = tf.expand_dims(series, axis=-1)
ds = tf.data.Dataset.from_tensor_slices(series)
ds = ds.window(window_size + 1, shift=1, drop_remainder=True)
ds = ds.flat_map(lambda w: w.batch(window_size + 1))
ds = ds.shuffle(shuffle_buffer)
ds = ds.map(lambda w: (w[:-1], w[1:]))
return ds.batch(batch_size).prefetch(1)
In [0]:
for X_batch, Y_batch in seq2seq_window_dataset(tf.range(10), 3,
batch_size=1):
print("X:", X_batch.numpy())
print("Y:", Y_batch.numpy())
In [0]:
keras.backend.clear_session()
tf.random.set_seed(42)
np.random.seed(42)
window_size = 30
train_set = seq2seq_window_dataset(x_train, window_size,
batch_size=128)
model = keras.models.Sequential([
keras.layers.SimpleRNN(100, return_sequences=True,
input_shape=[None, 1]),
keras.layers.SimpleRNN(100, return_sequences=True),
keras.layers.Dense(1),
keras.layers.Lambda(lambda x: x * 200)
])
lr_schedule = keras.callbacks.LearningRateScheduler(
lambda epoch: 1e-7 * 10**(epoch / 30))
optimizer = keras.optimizers.SGD(lr=1e-7, momentum=0.9)
model.compile(loss=keras.losses.Huber(),
optimizer=optimizer,
metrics=["mae"])
history = model.fit(train_set, epochs=100, callbacks=[lr_schedule])
In [0]:
plt.semilogx(history.history["lr"], history.history["loss"])
plt.axis([1e-7, 1e-4, 0, 30])
In [0]:
keras.backend.clear_session()
tf.random.set_seed(42)
np.random.seed(42)
window_size = 30
train_set = seq2seq_window_dataset(x_train, window_size,
batch_size=128)
valid_set = seq2seq_window_dataset(x_valid, window_size,
batch_size=128)
model = keras.models.Sequential([
keras.layers.SimpleRNN(100, return_sequences=True,
input_shape=[None, 1]),
keras.layers.SimpleRNN(100, return_sequences=True),
keras.layers.Dense(1),
keras.layers.Lambda(lambda x: x * 200.0)
])
optimizer = keras.optimizers.SGD(lr=1e-6, momentum=0.9)
model.compile(loss=keras.losses.Huber(),
optimizer=optimizer,
metrics=["mae"])
early_stopping = keras.callbacks.EarlyStopping(patience=10)
model.fit(train_set, epochs=500,
validation_data=valid_set,
callbacks=[early_stopping])
In [0]:
rnn_forecast = model_forecast(model, series[..., np.newaxis], window_size)
rnn_forecast = rnn_forecast[split_time - window_size:-1, -1, 0]
In [0]:
plt.figure(figsize=(10, 6))
plot_series(time_valid, x_valid)
plot_series(time_valid, rnn_forecast)
In [0]:
keras.metrics.mean_absolute_error(x_valid, rnn_forecast).numpy()