49 lines
2.1 KiB
Plaintext
49 lines
2.1 KiB
Plaintext
[[creating-df-kib]]
|
|
== Creating {dataframe-transforms}
|
|
|
|
beta[]
|
|
|
|
You can create {stack-ov}/ml-dataframes.html[{dataframe-transforms}] in the
|
|
{kib} Machine Learning application.
|
|
|
|
[role="screenshot"]
|
|
image::ml/images/ml-definepivot.jpg["Defining a {dataframe} pivot"]
|
|
|
|
Select the index pattern or saved search you want to transform. To pivot your
|
|
data, you must group the data by at least one field and apply at least one
|
|
aggregation. The {dataframe} pivot preview on the right side provides a visual
|
|
verification.
|
|
|
|
Once you have created the pivot, add a job ID and define the index for the
|
|
transformed data (_target index_). If the target index does not exist, it will be
|
|
created automatically. You can optionally select to create a {kib} index pattern
|
|
for the target index. At the end of the process, a {dataframe} job is created as
|
|
a result.
|
|
|
|
[role="screenshot"]
|
|
image::ml/images/ml-jobid.jpg["Job ID and target index"]
|
|
|
|
After you create {dataframe} jobs, you can start, stop, and delete them
|
|
and explore their progress and statistics from the jobs list.
|
|
|
|
For a more detailed example of using {dataframes} with the {kib} sample data,
|
|
see {stack-ov}/ecommerce-dataframes.html[Transforming your data].
|
|
|
|
[NOTE]
|
|
===============================
|
|
If {stack} {security-features} are enabled, you must have appropriate authority
|
|
to work with {dataframes}. For example, there are built-in
|
|
`data_frame_transforms_admin` and `data_frame_transforms_user` roles that have
|
|
`manage_data_frame_transforms` and `monitor_data_frame_transforms` cluster
|
|
privileges respectively. See
|
|
{stack-ov}/built-in-roles.html[Built-in roles] and
|
|
{stack-ov}/security-privileges.html[Security privileges].
|
|
|
|
Depending on what tasks you perform, you might require additional privileges.
|
|
For example, to create a {dataframe-transform} and generate a new target index,
|
|
you need `manage_data_frame_transforms` cluster privileges, `read` and
|
|
`view_index_metadata` privileges on the source index, and `read`, `create_index`,
|
|
and `index` privileges on the target index. For more information, see the
|
|
authorization details for each {ref}/data-frame-apis.html[{dataframe} API].
|
|
|
|
=============================== |