Summary
This chapter presented two case studies on how to apply the data science process with LightGBM. The data science life cycle and the typical constituent steps were discussed in detail.
A case study involving wind turbine power generation was presented as an example of approaching a data problem while working through the life cycle. Feature engineering and how to handle outliers were discussed in detail. An example exploratory data analysis was performed with samples given for visualization. Model training and tuning were shown alongside a basic script for exporting and using the model as a program.
A second case study involving multi-class credit score classification was also presented. The data science process was again followed, with particular attention given to data cleaning and class imbalance problems in the dataset.
The next chapter discusses the AutoML framework FLAML and introduces the concept of machine learning pipelines.
