Summary

Deep learning is a massive topic, with entire books dedicated to it. This chapter on LSTM units ends our adventures into this topic, but I hope your interest has been piqued and that you leave feeling ready to experiment with these types of models on your own.

By completing this chapter, you should be able to reshape your time series data in ways that suit your model requirements according to your domain expertise. You’ll find that architecture design is one of the hardest parts of forecasting with deep learning.

In Chapter 11, Anomaly Detection – Predicting Failure with No Failure Examples, you’ll learn how to use some of the forecasting techniques we’ve learned so far to detect anomalous data points or data drift for use cases such as machine maintenance.

Questions

1. How many season patterns can the LSTM model handle?
   1. One
   2. Two
   3. Any number
2. An LSTM unit is comprised of multiple neural layers. (True or False)
   1. True
   2. False
3. Which type...