Case study – using FLAML with LightGBM
We will use the Wind Turbine dataset from the previous chapter for the case study. The dataset is cleaned as before, imputing missing values and capping outliers to appropriate ranges. However, we take a different approach to feature engineering. To further explore AutoML, we use an open source framework called Featuretools.
Feature engineering
Featuretools (https://featuretools.alteryx.com/en/stable/#) is an open source framework for automated feature engineering. Specifically, Featuretools is well suited to transforming relational datasets and temporal data.
As discussed in the previous section, automated feature engineering tools typically use combinatorial transformations of features to generate new features for the dataset. Featuretools supports feature transformations through their Deep Feature Synthesis (DFS) process.
As an example, consider a dataset of online customer web sessions. Typical features that could be useful...
