In this chapter, we considered how to load data from CSV,  JSON, and HDF5 formats. We saw how to convert the loaded data in the objects suitable to use in different machine learning frameworks. We used the libraries' APIs to convert raw C++ arrays into matrices and higher-level datasets' objects for machine learning algorithms. We looked at how to load and process images with the OpenCV and Dlib libraries. We became familiar with the data normalization process, which is very important for many machine learning algorithms. Also, we saw which normalization techniques are available in machine learning libraries, and we implemented some normalization approaches with linear algebra functions from the Eigen library.

In the following chapter, we will see how to measure a model's performance on different types of data. We will look at special techniques that help us to understand how the model describes the training dataset well and how it performs on new data. Also, we will learn different types of parameters machine learning models depend on, and see how to select the best combination of them to improve the model's performance.