In this post, we will learn how to use DeepAR to forecast multiple time series using GluonTS in Python. DeepAR is a deep learning algorithm based on recurrent neural networks designed specifically for time series forecasting. It works by learning a model based on all the time series data, instead of creating a separate model for each one. In my experience, this often works better than creating a separate model for each time series....
Building an LSTM from scratch is very laborious and prone to mistakes, but we have libraries that will help us harness its power without having to spend hours debugging. We will use the NeuralForecast library which, in addition to implementing the most common structures of recurrent neural networks, also implements automatic hyperparameter tuning. Table of Contents How Does an LSTM Work? When to Use an LSTM How to Install NeuralForecast With and Without GPU Support How To Prepare Time Series Data For The LSTM Time Series Validation Split Automatic Time Series Modeling With LSTM NeuralForecast’s LSTM Architecture Automatic Hyperparameter Tuning input_size_multiplier encoder_hidden_size encoder_n_layers decoder_hidden_size learning_rate max_steps Training Code For LSTM In Python Hyperparameter Tuning Report Adding External Variables To The LSTM Custom PyTorch Loss Function (WMAPE) {#wmape-loss} Frequently Asked Questions How To Train The LSTM With Multiple SKUs?...