Creating Machine Learning Datasets Using SQL
Renée M. P. Teate · 2021
This chapter discusses the development of datasets for two types of algorithms: classification and time series models. Each type of model requires a different type of dataset, and the chapter reviews some approaches for preparing datasets for these two common types of models using Structured Query Language (SQL). Binary classification algorithms categorize inputs into one of two outcomes, which are determined according to the purpose of the model and the available data. Datasets that could be used as inputs to train time series models and binary classification models is built. Some people do feature engineering in their model-building script or other software. Tools like the pandas package in Python do make certain types of feature engineering straightforward to include in machine learning script. Some types of summarization are more efficient to do in SQL at the point of data extraction from the database than in other coding environments.