Research consistently shows that machine learning and data science practitioners spend most of their time manipulating data and preparing it for analysis. Indeed, many find it the most tedious and least enjoyable part of their work. Numerous companies are offering solutions to the problem but, in my opinion, results at this point are varied. Therefore, in this first chapter, I shall endeavor to provide a way of tackling the problem that will ease the burden of getting your data ready for machine learning. The methodology introduced in this chapter will serve as the foundation for data preparation and for understanding many of the subsequent chapters. I propose that once you become comfortable with this tried and true process, it may very well become your favorite part of machine learning—as it is for me.
The following are the topics that we'll cover in this chapter:
- Overview
- Reading the data
- Handling duplicate observations
- Descriptive statistics
- Exploring categorical variables
- Handling missing values
- Zero and near-zero variance features
- Treating the data
- Correlation and linearity
