chapter one
1 Understanding machine learning and deep learning with tabular data
This chapter covers
- What tabular data is
- Why tabular data matters
- The distinction between deep learning and non-deep learning approaches to tabular data
- What people think about using deep learning with tabular data
- Characteristics of tabular data that distinguish it from other kinds of data, like image, sound or text data
Tabular data is central to our modern lives and, for most of us, to our work lives. Tabular data exists in spreadsheets, as CSV files, and in the tables of relational databases. Tabular data populates analysis and reports but it is also often the fuel for training machine learning models. If we can determine how to efficiently train such models with tabular data, we’ll be able to successfully solve many useful problems, such as predicting stock requirements in retail outlets or predicting the price of market commodities. To train such models efficiently, we need to pick the right approach, both in terms of the model we choose for a particular problem and in terms of how we handle and transform the data.