Tensorflow sketchbook

dataset_timeserie_window.py

import tensorflow as tf

window_size = 2
label_size = 1

inputs = tf.random.stateless_uniform(shape=(10, 3), seed=(2, 3))

dataset = tf.data.Dataset.from_tensor_slices(inputs)
dataset = dataset.window(window_size + label_size, shift=label_size, drop_remainder=True)
dataset = dataset.flat_map(lambda window: window.batch(window_size + label_size))
dataset = dataset.map(lambda window: (window[:-label_size], window[-label_size]))

for x, y in dataset:
  print(x.numpy(), y.numpy())

fast_fourier_transform.py

import tensorflow as tf

fft = tf.signal.rfft(df['T (degC)'])
f_per_dataset = np.arange(0, len(fft))

n_samples_h = len(df['T (degC)'])
hours_per_year = 24*365.2524
years_per_dataset = n_samples_h/(hours_per_year)

f_per_year = f_per_dataset/years_per_dataset
plt.step(f_per_year, np.abs(fft))
plt.xscale('log')
plt.ylim(0, 400000)
plt.xlim([0.1, max(plt.xlim())])
plt.xticks([1, 365.2524], labels=['1/Year', '1/day'])
_ = plt.xlabel('Frequency (log scale)')

plot_keras_history.py

import pandas as pd

metrics = pd.DataFrame(history.history)
metrics.plot(xlabel="epochs", subplots=True, legend=True)

sliding_window_predicton.py

import numpy as np
import matplotlib.pyplot as plt

window_size = 2
split_time = 1000

forecast = range(len(series) - window_size)
forecast = map(lambda x: model.predict(series[x:x + window_size][np.newaxis]), forecast)
forecast = np.fromiter(forecast, dtype=np.float64)

forecast = forecast[split_time-window_size:]
forecast = np.array(forecast)[:, 0, 0]

plt.plot(time, forecast)
plt.xlabel("Time")
plt.ylabel("Value")
plt.grid(True)

timeserie_window.py

import tensorflow as tf

window_size = 2
label_size = 1

inputs = tf.random.stateless_uniform(shape=(10, 3), seed=(2, 3))
labels = inputs[:, 2]

inputs = tf.keras.utils.timeseries_dataset_from_array(inputs, None, sequence_length=window_size, sequence_stride=window_size+label_size)
labels = tf.keras.utils.timeseries_dataset_from_array(labels, None, sequence_length=label_size, sequence_stride=window_size+label_size, start_index=window_size)

dataset = tf.data.Dataset.zip((inputs, labels))

for x, y in dataset:
  print(x.numpy(), y.numpy())