The Book of Alternative Data

CHAPTER 8Missing Data: Case Studies

8.1. INTRODUCTION

In this chapter we will present real-world case studies on imputing missing values based on multivariate time series of credit default swap (CDS) data and satellite images. But before delving into it, let's start by introducing some notation.¹

For the case study in this section, we use a description in terms of a standard data matrix with observations and features. This means that has observations along the first index (rows) and different features along the second index (columns). Since we are dealing with multivariate time series, corresponds to the number of time series components and time stamps increase along the columns. It is noteworthy that a lot of what we discuss also applies to data in different formats such as heterogeneous data with different features. Image data, for example, can be represented with either pixel ...