Introduction

The conclusions drawn from data analysis are only as robust as the quality of the data itself. After obtaining raw text, the next natural step is to validate and clean it carefully. Even the slightest bias may risk the integrity of the results. Therefore, we must take great precautionary measures, which involve thorough inspection, to ensure sanity checks are performed on our data before we begin to understand it. This section should be the starting point for cleaning data in Haskell.

Real-world data often has an impurity that needs to be addressed before it can be processed. For example, extraneous whitespaces or punctuation could clutter data, making it difficult to parse. Duplication and data conflicts are another area of unintended consequences of reading real-world data. Sometimes it's just reassuring to know that data makes sense by conducting sanity checks. Some examples of sanity checks include matching regular expressions as well as detecting outliers by establishing a measure of distance. In this chapter, we will cover each of these topics.