TCNForecaster

---

TCNForecaster

Introduction

Temporal Convolutional Networks (TCN) is a neural network that use convolutional architecture rather than recurrent networks. It supports multi-step and multi-variant cases. Causal Convolutions enables large scale parallel computing which makes TCN has less inference time than RNN based model such as LSTM.

Method

Arguments

• past_seq_len: Specify the history time steps (i.e. lookback).
• future_seq_len: Specify the output time steps (i.e. horizon).
• input_feature_num: Specify the feature dimension.
• output_feature_num: Specify the output dimension.
• num_channels: Specify the convolutional layer filter number in TCN's encoder. This value defaults to [30]*8.
• kernel_size: Specify convolutional layer filter height in TCN's encoder. This value defaults to 7.
• dropout: Specify the dropout close possibility (i.e. the close possibility to a neuron). This value defaults to 0.2.
• optimizer: Specify the optimizer used for training. This value defaults to "Adam".
• loss: Specify the loss function used for training. This value defaults to "mse". You can choose from "mse", "mae" and "huber_loss".
• lr: Specify the learning rate. This value defaults to 0.001.

__init__

TCNForecaster(past_seq_len,
              future_seq_len,
              input_feature_num,
              output_feature_num,
              num_channels=[30]*8,
              kernel_size=7,
              dropout=0.2,
              optimizer="Adam",
              loss="mse",
              lr=0.001)

fit

fit(x, y, validation_data=None, epochs=1, metric="mse", batch_size=32)

• x: A numpy array with shape (num_samples, lookback, feature_dim). lookback and feature_dim should be the same as past_seq_len and input_feature_num.
• y: A numpy array with shape (num_samples, horizon, target_dim). horizon and target_dim should be the same as future_seq_len and output_feature_num.
• validation_data: A tuple (x_valid, y_valid) as validation data. Default to None.
• epochs: Number of epochs you want to train.
• batch_size: Number of batch size you want to train.
• metric: The metric for training data.

evaluate

evaluate(x, y, metrics=['mse'], multioutput="raw_values")

• x: A numpy array with shape (num_samples, lookback, feature_dim).
• y: A numpy array with shape (num_samples, horizon, target_dim).
• metrics: A list contains metrics for test/valid data.
• multioutput: Defines aggregating of multiple output values. String in ['raw_values', 'uniform_average']. The value defaults to 'raw_values'.

evaluate_with_onnx

evaluate_with_onnx(x, y, metrics=['mse'], dirname=None, multioutput="raw_values")

• x: A numpy array with shape (num_samples, lookback, feature_dim).
• y: A numpy array with shape (num_samples, horizon, target_dim).
• metrics: A list contains metrics for test/valid data.
• dirname: The directory to save onnx model file. This value defaults to None for no saving file.
• multioutput: Defines aggregating of multiple output values. String in ['raw_values', 'uniform_average']. The value defaults to 'raw_values'.

predict

predict(x)

• x: A numpy array with shape (num_samples, lookback, feature_dim).

predict_with_onnx

predict_with_onnx(x, dirname=None)

• x: A numpy array with shape (num_samples, lookback, feature_dim).
• dirname: The directory to save onnx model file. This value defaults to None for no saving file.

save

save(checkpoint_file)

• checkpoint_file: The location you want to save the forecaster.

restore

restore(checkpoint_file)

• checkpoint_file: The checkpoint file location you want to load the forecaster.