Using features from graph algorithms in a scikit-learn pipeline

Now we have all the necessary knowledge to actually use graphs for ML. In this section, we are going to wrap everything up using the GDS Python client to create features and extract data into a dataframe that can be fed into a scikit-learn model training pipeline.

But before we get to this, let me give you an overview of the ML possibilities with graphs.

Machine learning tasks with graphs

In general, ML comprises several types of tasks on various kinds of objects: from sales predictions with time series analysis to patient diagnosis thanks to medical imagery to text translation in many languages with natural language processing (NLP), ML has proven its usefulness in many situations.

In each of these cases, you have to build a dataset made of observations (usually, the rows). Each observation has a certain number of characteristics or features (that is, the columns of your dataset). Depending on the task, you...