Book Image

SQL for Data Analytics - Third Edition

By : Jun Shan, Matt Goldwasser, Upom Malik, Benjamin Johnston
Book Image

SQL for Data Analytics - Third Edition

By: Jun Shan, Matt Goldwasser, Upom Malik, Benjamin Johnston

Overview of this book

Every day, businesses operate around the clock, and a huge amount of data is generated at a rapid pace. This book helps you analyze this data and identify key patterns and behaviors that can help you and your business understand your customers at a deep, fundamental level. SQL for Data Analytics, Third Edition is a great way to get started with data analysis, showing how to effectively sort and process information from raw data, even without any prior experience. You will begin by learning how to form hypotheses and generate descriptive statistics that can provide key insights into your existing data. As you progress, you will learn how to write SQL queries to aggregate, calculate, and combine SQL data from sources outside of your current dataset. You will also discover how to work with advanced data types, like JSON. By exploring advanced techniques, such as geospatial analysis and text analysis, you will be able to understand your business at a deeper level. Finally, the book lets you in on the secret to getting information faster and more effectively by using advanced techniques like profiling and automation. By the end of this book, you will be proficient in the efficient application of SQL techniques in everyday business scenarios and looking at data with the critical eye of analytics professional.
Table of Contents (11 chapters)
9. Using SQL to Uncover the Truth: A Case Study

Working with Missing Data

In all the examples so far, you have been dealing with datasets that are clean and easy to decipher. However, datasets in real world are more complicated than these. One of the many problems you may have to deal with when working with datasets is missing values.

You will further learn the specifics of preparing data in Chapter 3, SQL for Data Preparation. However, in this section, you will learn several strategies that you can use to handle missing data. Some of your strategies include the following:

  • Deleting rows: If a very small number of rows (that is, less than 5% of your dataset) is missing data, then the simplest solution may be to just delete the data points from your set. This would not impact your results too much.
  • Mean/median/mode imputation: If 5% to 25% of your data for a variable is missing, another option is to take the mean, median, or mode of that column and fill in the blanks with that value. It may provide a...