Setting up your workspace

Before we can start, you will need to set up your workspace. The examples in this book are all meant to run in a Jupyter notebook. Jupyter notebooks are an interactive development environment mostly used for data-science applications and are considered the go-to environment to build data-driven applications in.

You can run Jupyter notebooks either on your local machine, on a server in the cloud, or on a website such as Kaggle.

Note

Note: All code examples for this book can be found here: https://github.com/PacktPublishing/Machine-Learning-for-Finance and for chapter 1 refer the following link: https://www.kaggle.com/jannesklaas/machine-learning-for-finance-chapter-1-code.

Deep learning is computer intensive, and the data used in the examples throughout this book are frequently over a gigabyte in size. It can be accelerated by the use of Graphics Processing Units (GPUs), which were invented for rendering video and games. If you have a GPU enabled computer, you can run the examples locally. If you do not have such a machine, it is recommended to use a service such as Kaggle kernels.

Learning deep learning used to be an expensive endeavor because GPUs are an expensive piece of hardware. While there are cheaper options available, a powerful GPU can cost up to $10,000 if you buy it and about $0.80 an hour to rent it in the cloud.

If you have many, long-running training jobs, it might be worth considering building a "deep learning" box, a desktop computer with a GPU. There are countless tutorials for this online and a decent box can be assembled for as little as a few hundred dollars all the way to $5,000.

The examples in this book can all be run on Kaggle for free, though. In fact, they have been developed using this site.