Introducing MLflow

Simply put, MLflow is there to simplify the model development lifecycle. A lot of the data scientist's time is spent finding the right algorithms with the right hyperparameters for the given dataset. As a data scientist, you experiment with different combinations of parameters and algorithms, then review and compare the results to make the right choice. MLflow allows you to record, track, and compare these parameters, their results, and associated metrics. The component of MLflow that captures the details of each of your experiments is called the tracking server. The tracking server captures the environment details of your notebook, such as the Python libraries and their versions, and the artifacts generated by your experiment.

The tracking server allows you to compare the data captured between different runs of an experiment, such as the performance metrics (for example, accuracy) alongside the hyperparameters used. You can also share this data with your...