Data exploration

If the data you need to analyze is already well-specified in a schema, such as if it was exported from another database, then you can usually proceed to defining tables and loading the data. Often, however, your initial data is less clear or is in a format that does not require a consistent schema, like JSON. Before you can define database tables for your data, you must explore that data in more detail and make decisions about how to represent it in tables.

During data exploration, you develop a schema and decide what kind of table to use in your production database. In some cases, you might choose to use external tables to describe data, and in other cases you might choose to load data into the database. Either way, over time, you might need to make changes if the schema evolves. For example, new data might add columns or change the data type of existing columns. You should expect to revisit your data schema over time. When you do, use the techniques described in the following sections.