An Introduction to Machine Learning in Quantitative Finance

The following sections are included:

• Dedication
• Preface
• About the Authors
• Acknowledgments
• Disclaimer
• Contents
• Listings

The following sections are included:

• Big Data Era
• Machine Learning
• Quantitative Finance
  • Challenges of financial data
  • Recent development of machine learning for financial applications
  • The future of quantitative finance
• Next Generation of Talents in Quantitative Finance
• Outline of the Book
• Useful Resources
  • Python libraries
  • Books and other online reading materials
• Go Beyond This Book

Supervised learning is the machine learning task of inferring a function that maps an input to an output based on example input–output pairs. Depending on whether the output is a continuous variable or a categorical variable, the supervised learning can be further divided into two types:

• Regression;
• Classification…

In this chapter, we focus on linear regression and explain how it fits in the general framework of supervised learning. We start with the simplest linear model—Ordinary Least Squares (OLS)—and then discuss the natural extension of OLS, i.e., the linear model with regularization.

The following sections are included:

• Introduction
• Decision Tree
  • Tree structure
  • Model
  • Regression decision tree
  • Pruning
  • Feature importance
• Random Forest
• Gradient Boosting Decision Tree
• Numerical Example: Iris Dataset
  • Decision tree implementation
  • Random forest implementation
  • GBDT implementation
  • Comparison of three tree-based methods
• Exercises

In March 2016, AlphaGo beat Lee Sedol by 4–1 in five games. It is the first time that a computer Go program has defeated a human 9 dan world champion without handicaps. The deep neural network is one of the techniques that AlphaGo used. AlphaGo’s victory sparked the discussion on artificial intelligence and made the concept of deep learning well known and accessible to ordinary people all over the world…

In the supervised learning chapters, we have seen machine learning algorithms based on the assumptions or model of data. In this chapter, we focus on cluster analysis as another type of algorithm aimed at exploring the general and natural characteristics of multidimensional data without pre-specified data model assumptions.

In this chapter, we introduce principal component analysis (PCA) as another unsupervised learning method. PCA transforms the original data into another data representation where information is ordered in orthogonal dimensions. There are many applications, and in the following section, we focus on using PCA to reduce the number of data dimensions with information loss under control.

The following sections are included:

• Introduction
• Recurrent Reinforcement Learning
• Link between RRL and RNN
• Numerical Example: Algorithmic Trading
• Exercises

In this last chapter, we aim to equip readers with hands-on experience in the application of machine learning to real-world data challenges. The pipelines of machine learning applications include the main steps: (1) Pre-process the data; (2) Extract features; (3) Train a base model; (4) Parameter tuning; (5) Ensemble base models; and (6) Make final predictions (Figure 9.1)…

The following sections are included:

• Bibliography
• Index