AutoML with LightGBM and FLAML
In the previous chapter, we discussed two case studies that showed end-to-end examples of how to approach data science problems. Of the steps involved in the typical data science life cycle, often, the most time-consuming tasks are preparing the data, finding the correct models, and tuning the models.
This chapter looks at the concept of automated machine learning. Automated machine learning systems seek to automate some or all parts of the machine learning life cycle. We will look at FLAML, a library that automates the process’s model selection and tuning steps using efficient hyperparameter optimization algorithms.
Lastly, we will present a case study using FLAML and another open source tool called Featuretools. Practical usage of FLAML will be discussed and shown. We will also show FLAML’s zero-shot AutoML functionality, which bypasses tuning altogether.
The main topics of this chapter are as follows:
- An introduction...
