<aside> đź’ˇ Adding a dataset requires a connected database. For instructions on connecting a database or data warehouse, click here!
</aside>
AirOps allows you to create datasets using multiple methods, each of which has its own benefits and considerations.
AirOps allows users to create a dataset using standard SQL syntax. This is a good option for users who are comfortable with SQL and have a specific query in mind. To create a dataset using SQL, users simply need to enter their query into the SQL editor and click "Save".
AirOps also offers a natural language interface for creating datasets. This allows users to create a dataset by typing out their query in plain English. This is a good option for users who are not familiar with SQL or want a more intuitive way to create their dataset.
The “freeform” option allows users to enter any question, statement, or command to generate a SQL query. Alternatively, the “Structured” option can be used for more explicit requests (e.g. choosing specific columns from a table).
After the request has been entered, AirOps can attempt to automatically identify the correct table(s) using the “🔮 Suggest Tables” option, or the table can be selected from the schema browser on the left side.
The visual query builder allows users to create a new dataset from an existing dataset, table, or model. Users can select a dataset from the left-hand dataset list. After that, the dataset can be refined using the visual query builder tools (e.g. filter, sort, add/remove columns) or by using the “Refine with AI” option.
Refine with AI allows users to give requests in plain english to refine the dataset to the exact table they want to use.
AirOps also supports creating datasets from tables that already exist in the data warehouse, or tables that were created from external sources such as dbt. This allows users to use pre-existing queries and models from these sources in their AirOps datasets. To create a dataset from an existing table, users need to select the "Existing Table" option in the dataset creation page and follow the prompts to connect and select the desired table, view, or model.