Extract Transform Load
ETL (Extract, Transform, Load) takes data from a source, reshapes it in a series of small steps, and stores it at a destination.
Apprentice · General · Updated
In data science, we often work with large data sets to move them from one database to another. This process is called ETL (Extract Transform Load).
We Extract data from the source, Transform the data to make it match the format we need, and Load it into our destination.
If the data format at the source exactly matches the data format at the destination, then no transformation needs to be performed.
In reality, though, data at the source almost never matches the format that we need at the destination. So a data flow would look more like this.
Notice that there is not just one single transformation. Instead each transformation performs a single task as the data flows through the stream, then passes the data onto the next step.
All ETL processes begin with a source (data at rest), move through one or more transformations in a stream (data in transit), and then store the result in a destination (data at rest).
Key terms
- ETL