Book Image

Data Exploration and Preparation with BigQuery

By : Mike Kahn
Book Image

Data Exploration and Preparation with BigQuery

By: Mike Kahn

Overview of this book

Data professionals encounter a multitude of challenges such as handling large volumes of data, dealing with data silos, and the lack of appropriate tools. Datasets often arrive in different conditions and formats, demanding considerable time from analysts, engineers, and scientists to process and uncover insights. The complexity of the data life cycle often hinders teams and organizations from extracting the desired value from their data assets. Data Exploration and Preparation with BigQuery offers a holistic solution to these challenges. The book begins with the basics of BigQuery while covering the fundamentals of data exploration and preparation. It then progresses to demonstrate how to use BigQuery for these tasks and explores the array of big data tools at your disposal within the Google Cloud ecosystem. The book doesn’t merely offer theoretical insights; it’s a hands-on companion that walks you through properly structuring your tables for query efficiency and ensures adherence to data preparation best practices. You’ll also learn when to use Dataflow, BigQuery, and Dataprep for ETL and ELT workflows. The book will skillfully guide you through various case studies, demonstrating how BigQuery can be used to solve real-world data problems. By the end of this book, you’ll have mastered the use of SQL to explore and prepare datasets in BigQuery, unlocking deeper insights from data.
Table of Contents (21 chapters)
Free Chapter
1
Part 1: Introduction to BigQuery
4
Part 2: Data Exploration with BigQuery
10
Part 3: Data Preparation with BigQuery
14
Part 4: Hands-On and Conclusion

Data preparation

By examining Figure 12.7, you can see that many of the columns in our dataset are of the data type FLOAT. While FLOAT is a legacy SQL data type, the GoogleSQL modern datatype is FLOAT64. FLOAT64 provides higher precision than FLOAT as FLOAT64 uses 64 bits to represent floating-point numbers, while FLOAT uses 32 bits.

For this hands-on example, we will leave most of the FLOAT and other data types and only modify the TIME column. We will be able to gain the insights we need from our dataset by leaving most of the data types as they are.

Figure 12.8 – Previewing our GPS data set to examine the TIME column

Note in Figure 12.7 that the TIME column is FLOAT, has a decimal point, and is not very readable. Upon loading, the TIME column is formatted YYYYMMDDHHMMSS (14 digits). To prepare our dataset, we will convert this column into YYYY-MM-DD format (8 digits).

To convert the TIME column, we will use BigQuery Data Definition Language...