Introduction

Having gone through a fairly involved analysis in the previous chapter, you should now be feeling comfortable using Jupyter Notebooks to work with data. In addition to data exploration and visualization, our analysis included a couple of relatively simple modeling problems, where we trained linear regression models. These lines of best fit were very easy to create because only two dimensions were involved and the data was very clean.

As we will see in later chapters, training more advanced models (such as decision trees) can be just as easy because of the simplicity of open source software such as scikit-learn. The work involved in preparing data, however, can be significantly more difficult, depending on the details of the relevant datasets.

The quality of training data is very important for creating a model that will generalize well to future samples. For example, errors in your training dataset will cause the model to learn patterns that don't reflect the...