Create a dataset¶
A dataset, which is the basic unit of data preparation, refers to an entity subject to data operations. Datasets are either imported datasets and wrangled datasets.
Imported Dataset: A source data entity before the implementation of transformation rules
Wrangled dataset: A data entity subject to analysis following the implementation of transformation rules
A wrangled dataset is created during the dataflow setting process, which defines transformation rules, while an imported dataset is created during this dataset creation procedure.
The Dataflow menu can be accessed under MANAGEMENT > Data Preparation > Dataset on the left-hand panel of the main screen.
Next, on the upper right of the dataset page, click the + Generate new dataset button to create a new dataset.
In the dataset creation page, select the dataset type.
My file: Create a dataset by opening the user’s local file or via a URI (upcoming feature) (See Create a dataset from a file for a detailed procedure).
Database: Create a dataset using external database access information and queries (See Create a dataset from a database for a detailed procedure).
Staging DB: Create a dataset from the staging DB built in Metatron (See Create a dataset from staging DB for a detailed procedure).
Note
The Staging DB is an in-cluster database that stores data temporarily in order to facilitate data loading. Hive is generally used for it.
Create a dataset from a file¶
Create a dataset by opening the user’s local file or via a URI (upcoming feature).
On the data type selection page, select My File.
Select a file to be used as a data source from your local PC. You can click the Import button to select a file, or drag and drop the file into the box. Once a file is selected, click Next.
Check the grid of the uploaded file, and designate a column delimiter. Proceed if the data is successfully displayed.
Enter the Name and Description of the dataset, and click the Done button.
Once the dataset is created, the dataset list is displayed. You can check that the list contains the newly created dataset.
Create a dataset from a database¶
Create a dataset using external database access information and queries.
To create a dataset from a database, you should first create a data connection. See Create a data connection for a detailed procedure.
After establishing the data connection, go to MANAGEMENT > Data Preparation > Dataset > + Generate new dataset.
On the data type selection page, select Database.
Select the data connection, and press the Test button to check that the connection is valid.
Select the data. You can either select a table from the connected database, or write a query yourself.
Table: Select a database and a table to display the table’s data. Once the data being ingested has been displayed, confirm the data and click Next.
Query: Write a query to import the data you want, and click Run to display the data in the lower section. Confirm the data and click Next.
Enter the Name and Description of the dataset, and click the Done button.
Once the dataset is created, the dataset list is displayed. You can check that the list contains the newly created dataset.
Create a dataset from staging DB¶
Create a dataset from the staging DB built in Metatron.
The creation of a staging DB dataset is the same as dataset creation from a database, but does not involve the selection of a data connection.
On the data type selection page, select Staging DB.
Select the data. You can either select a table from the connected database, or write a query yourself.
Table: Select a database and a table to display the table’s data. Once the data being ingested has been displayed, confirm the data and click Next.
Query: Write a query to import the data you want, and click Run to display the data in the lower section. Confirm the data and click Next.
Enter the Name and Description of the dataset, and click the Done button.
Once the dataset is created, the dataset list is displayed. You can check that the list contains the newly created dataset.
