Why use dbt

Although the strongest evidence exists for DBT as a treatment for people with borderline personality disorder, DBT has been found to be effective for a wide variety of mental health conditions. Conditions for which standard or adapted versions of DBT have been found to be effective in at least one randomized controlled trial are listed below Note: DBT has been evaluated for many other conditions in non-RCT research.

Although DBT was originally developed and researched in the United States, it has since been evaluated in many randomized controlled trials RCTs conducted in other countries, including:. These international RCTs have found that DBT is effective when implemented in other cultures and results are similar to those found in studies conducted in the United States.

DBT has been implemented across the world, making it one of the few evidence-based psychological treatments with a truly global reach see map. On this page: Why is DBT effective? The dbt data transformation tool, which is SQL-based, lets data professionals build data models iteratively and automate data transformation.

Users model their data using SQL SELECT statements, create relationships and dependencies between models, and then materialize those models as tables and views in a data warehouse. As it applies to data analytics, data modeling is the process of imposing order on data from diverse sources.

That means transforming tables from different schemas or even databases into a single view or table that describes a business system or process. The YAML file contains project configuration information. SQL has long been the language of choice for analysts developing transformations and performing analysis.

Using SQL makes for easy collaboration and a better understanding of your transformations. These packages contain pre-built SQL, drastically reducing the necessity to hand-code transformations.

Fivetran develops and maintains open-source dbt Packages which model data from single schema and multi-schema data sources. These Fivetran dbt Packages leverage data modeling principles and code modularity to effectively transform data to be used for exploratory, predictive, or prescriptive analytics. For more information on the Fivetran dbt packages and how to use them, you can refer to the dbt hub.

The YAML files are not only used to configure your dbt project, but are also used to curate a data dictionary of your source and transformed data as well as apply data integrity tests. If used effectively, the final result of your dbt project will have a definition of each raw table from the source and transformed table from your dimensional schema, as well as a description of each field within those tables.

Further, you can apply tests to certain fields ie. Models are modular, so you can reproduce a transformation across multiple projects. Any developer can tell you how valuable code reuse is in terms of saving time and effort. A project may also include test, macro, and documentation files. The dbt platform comprises two products. It takes a dbt project and a command and creates tables or views in your warehouse.

Some graphs are fairly straightforward. Other graphs are much more complicated. The following graph is from a venture-funded ecommerce company:. If statements, for loops, filters, macros, and more.

It turns out that SQL is far more powerful when you pair it with a fully-featured templating language. In addition to stock Jinja, dbt provides additional functions and variables within the Jinja context that allow users to express data transformation logic straightforwardly. This is an extremely useful pattern in practice: this model will execute different SQL depending on the current environment dev vs prod and state of the database.

For tables with a billion or more rows of data, adopting this pattern can significantly improve performance in both environments. Writing modular code like this allows users to automate complicated and time-consuming tasks that previously required manual effort. You can find the code here. To that end, dbt ships with a package manager.

Most analysts are used to being users of features, not developers of features. With dbt, analysts can do the same.

This past week I had to release a new version of my Stripe package when a data engineer at Buffer let me know in a Github issue that I had a bug.