Time-Series prediction for Univariate and Multivariate Sequences

Sequential data generation is an easy process. Other than language data, you simply have to observe an event happening periodically and note down occurrence of activities. Deep neural networks have a special class of nets for finding patterns from such sequences that could help to decouple some important information for decision making.

The DNNs specialised for processing sequences are covered under umbrella of Recurrent Neural Networks. They could be specialised for solving some of the common activities which we see these days as:-

• Named Entity recognition
• Machine Translation
• Question Answering
• Conversational bots
• and the list is endless……

In this particular blog post, we are going to learn

• how to create sequences using given dataset
• how to train models using dataset of sequences
• how to deal with univariate and multivariate dataset

Univariate and Multivariate dataset

Before talking much about code and process of training, we will see what is difference between uni- and multi- variate datasets. Univariate and multivariate represent two…