View Source AWS.Personalize (aws-elixir v0.13.3)
Amazon Personalize is a machine learning service that makes it easy to add individualized recommendations to customers.
Link to this section Summary
Functions
Creates a batch inference job.
Creates a batch segment job.
Creates a campaign that deploys a solution version.
Creates an empty dataset and adds it to the specified dataset group.
Creates a job that exports data from your dataset to an Amazon S3 bucket.
Creates an empty dataset group.
Creates a job that imports training data from your data source (an Amazon S3 bucket) to an Amazon Personalize dataset.
Creates an event tracker that you use when adding event data to a specified dataset group using the PutEvents API.
Creates a recommendation filter.
Creates a metric attribution.
Creates a recommender with the recipe (a Domain dataset group use case) you specify.
Creates an Amazon Personalize schema from the specified schema string.
Creates the configuration for training a model.
Trains or retrains an active solution in a Custom dataset group.
Removes a campaign by deleting the solution deployment.
Deletes a dataset.
Deletes a dataset group.
Deletes the event tracker.
Deletes a filter.
Deletes a metric attribution.
Deactivates and removes a recommender.
Deletes a schema.
Deletes all versions of a solution and the Solution
object itself.
Describes the given algorithm.
Gets the properties of a batch inference job including name, Amazon Resource Name (ARN), status, input and output configurations, and the ARN of the solution version used to generate the recommendations.
Gets the properties of a batch segment job including name, Amazon Resource Name (ARN), status, input and output configurations, and the ARN of the solution version used to generate segments.
Describes the given campaign, including its status.
Describes the given dataset.
Describes the dataset export job created by CreateDatasetExportJob, including the export job status.
Describes the given dataset group.
Describes the dataset import job created by CreateDatasetImportJob, including the import job status.
Describes an event tracker.
Describes the given feature transformation.
Describes a filter's properties.
Describes a metric attribution.
Describes a recipe.
Describes the given recommender, including its status.
Describes a schema.
Describes a solution.
Describes a specific version of a solution.
Gets the metrics for the specified solution version.
Gets a list of the batch inference jobs that have been performed off of a solution version.
Gets a list of the batch segment jobs that have been performed off of a solution version that you specify.
Returns a list of campaigns that use the given solution.
Returns a list of dataset export jobs that use the given dataset.
Returns a list of dataset groups.
Returns a list of dataset import jobs that use the given dataset.
Returns the list of datasets contained in the given dataset group.
Returns the list of event trackers associated with the account.
Lists all filters that belong to a given dataset group.
Lists the metrics for the metric attribution.
Lists metric attributions.
Returns a list of available recipes.
Returns a list of recommenders in a given Domain dataset group.
Returns the list of schemas associated with the account.
Returns a list of solution versions for the given solution.
Returns a list of solutions that use the given dataset group.
Get a list of tags attached to a resource.
Starts a recommender that is INACTIVE.
Stops a recommender that is ACTIVE.
Stops creating a solution version that is in a state of CREATE_PENDING or CREATE IN_PROGRESS.
Add a list of tags to a resource.
Remove tags that are attached to a resource.
Updates a campaign by either deploying a new solution or changing the value of
the campaign's minProvisionedTPS
parameter.
Updates a metric attribution.
Updates the recommender to modify the recommender configuration.
Link to this section Functions
Creates a batch inference job.
The operation can handle up to 50 million records and the input file must be in JSON format. For more information, see Creating a batch inference job.
Creates a batch segment job.
The operation can handle up to 50 million records and the input file must be in JSON format. For more information, see Getting batch recommendations and user segments.
Creates a campaign that deploys a solution version.
When a client calls the GetRecommendations and GetPersonalizedRanking APIs, a campaign is specified in the request.
minimum-provisioned-tps-and-auto-scaling
Minimum Provisioned TPS and Auto-Scaling
A transaction is a single GetRecommendations
or GetPersonalizedRanking
call.
Transactions per second (TPS) is the throughput and unit of billing for Amazon
Personalize. The minimum provisioned TPS (minProvisionedTPS
) specifies the
baseline throughput provisioned by Amazon Personalize, and thus, the minimum
billing charge.
If your TPS increases beyond minProvisionedTPS
, Amazon Personalize auto-scales
the provisioned capacity up and down, but never below minProvisionedTPS
.
There's a short time delay while the capacity is increased that might cause loss
of transactions.
The actual TPS used is calculated as the average requests/second within a
5-minute window. You pay for maximum of either the minimum provisioned TPS or
the actual TPS. We recommend starting with a low minProvisionedTPS
, track your
usage using Amazon CloudWatch metrics, and then increase the minProvisionedTPS
as necessary.
status
Status
A campaign can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
DELETE PENDING > DELETE IN_PROGRESS
To get the campaign status, call
DescribeCampaign. Wait until the status
of the campaign is ACTIVE
before asking the campaign
for recommendations.
related-apis
Related APIs
Creates an empty dataset and adds it to the specified dataset group.
Use CreateDatasetImportJob to import your training data to a dataset.
There are three types of datasets:
Interactions
Items
Users
Each dataset type has an associated schema with required field types. Only the
Interactions
dataset is required in order to train a model (also referred to
as creating a solution).
A dataset can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
DELETE PENDING > DELETE IN_PROGRESS
To get the status of the dataset, call DescribeDataset.
related-apis
Related APIs
Creates a job that exports data from your dataset to an Amazon S3 bucket.
To allow Amazon Personalize to export the training data, you must specify an
service-linked IAM role that gives Amazon Personalize PutObject
permissions
for your Amazon S3 bucket. For information, see Exporting a dataset in
the Amazon Personalize developer guide.
status
Status
A dataset export job can be in one of the following states:
- CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
To get the status of the export job, call
DescribeDatasetExportJob,
and specify the Amazon Resource Name (ARN) of the dataset export job. The
dataset export is complete when the status shows as ACTIVE. If the status shows
as CREATE FAILED, the response includes a failureReason
key, which describes
why the job failed.
Creates an empty dataset group.
A dataset group is a container for Amazon Personalize resources. A dataset group can contain at most three datasets, one for each type of dataset:
Interactions
Items
Users
A dataset group can be a Domain dataset group, where you specify a domain and use pre-configured resources like recommenders, or a Custom dataset group, where you use custom resources, such as a solution with a solution version, that you deploy with a campaign. If you start with a Domain dataset group, you can still add custom resources such as solutions and solution versions trained with recipes for custom use cases and deployed with campaigns.
A dataset group can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
DELETE PENDING
To get the status of the dataset group, call
DescribeDatasetGroup. If the status shows as CREATE FAILED, the response includes a failureReason
key, which describes why the creation failed.
You must wait until the status
of the dataset group is ACTIVE
before adding
a dataset to the group.
You can specify an Key Management Service (KMS) key to encrypt the datasets in the group. If you specify a KMS key, you must also include an Identity and Access Management (IAM) role that has permission to access the key.
apis-that-require-a-dataset-group-arn-in-the-request
APIs that require a dataset group ARN in the request
CreateEventTracker CreateSolution
related-apis
Related APIs
Creates a job that imports training data from your data source (an Amazon S3 bucket) to an Amazon Personalize dataset.
To allow Amazon Personalize to import the training data, you must specify an IAM service role that has permission to read from the data source, as Amazon Personalize makes a copy of your data and processes it internally. For information on granting access to your Amazon S3 bucket, see Giving Amazon Personalize Access to Amazon S3 Resources.
By default, a dataset import job replaces any existing data in the dataset that you imported in bulk. To add new records without replacing existing data, specify INCREMENTAL for the import mode in the CreateDatasetImportJob operation.
status
Status
A dataset import job can be in one of the following states:
- CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
To get the status of the import job, call
DescribeDatasetImportJob, providing the Amazon Resource Name (ARN) of the dataset import job. The dataset
import is complete when the status shows as ACTIVE. If the status shows as
CREATE FAILED, the response includes a failureReason
key, which describes why
the job failed.
Importing takes time. You must wait until the status shows as ACTIVE before training a model using the dataset.
related-apis
Related APIs
Creates an event tracker that you use when adding event data to a specified dataset group using the PutEvents API.
Only one event tracker can be associated with a dataset group. You will get an
error if you call CreateEventTracker
using the same dataset group as an
existing event tracker.
When you create an event tracker, the response includes a tracking ID, which you pass as a parameter when you use the PutEvents operation. Amazon Personalize then appends the event data to the Interactions dataset of the dataset group you specify in your event tracker.
The event tracker can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
DELETE PENDING > DELETE IN_PROGRESS
To get the status of the event tracker, call DescribeEventTracker. The event tracker must be in the ACTIVE state before using the tracking ID.
related-apis
Related APIs
Creates a recommendation filter.
For more information, see Filtering recommendations and user segments.
Creates a metric attribution.
A metric attribution creates reports on the data that you import into Amazon Personalize. Depending on how you imported the data, you can view reports in Amazon CloudWatch or Amazon S3. For more information, see Measuring impact of recommendations.
Creates a recommender with the recipe (a Domain dataset group use case) you specify.
You create recommenders for a Domain dataset group and specify the recommender's Amazon Resource Name (ARN) when you make a GetRecommendations request.
minimum-recommendation-requests-per-second
Minimum recommendation requests per second
When you create a recommender, you can configure the recommender's minimum
recommendation requests per second. The minimum recommendation requests per
second (minRecommendationRequestsPerSecond
) specifies the baseline
recommendation request throughput provisioned by Amazon Personalize. The default
minRecommendationRequestsPerSecond is 1
. A recommendation request is a single
GetRecommendations
operation. Request throughput is measured in requests per
second and Amazon Personalize uses your requests per second to derive your
requests per hour and the price of your recommender usage.
If your requests per second increases beyond
minRecommendationRequestsPerSecond
, Amazon Personalize auto-scales the
provisioned capacity up and down, but never below
minRecommendationRequestsPerSecond
. There's a short time delay while the
capacity is increased that might cause loss of requests.
Your bill is the greater of either the minimum requests per hour (based on
minRecommendationRequestsPerSecond) or the actual number of requests. The actual
request throughput used is calculated as the average requests/second within a
one-hour window. We recommend starting with the default
minRecommendationRequestsPerSecond
, track your usage using Amazon CloudWatch
metrics, and then increase the minRecommendationRequestsPerSecond
as
necessary.
status
Status
A recommender can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
STOP PENDING > STOP IN_PROGRESS > INACTIVE > START PENDING > START IN_PROGRESS > ACTIVE
DELETE PENDING > DELETE IN_PROGRESS
To get the recommender status, call DescribeRecommender.
Wait until the status
of the recommender is ACTIVE
before asking the
recommender for recommendations.
related-apis
Related APIs
Creates an Amazon Personalize schema from the specified schema string.
The schema you create must be in Avro JSON format.
Amazon Personalize recognizes three schema variants. Each schema is associated with a dataset type and has a set of required field and keywords. If you are creating a schema for a dataset in a Domain dataset group, you provide the domain of the Domain dataset group. You specify a schema when you call CreateDataset. ## Related APIs
Creates the configuration for training a model.
A trained model is known as a solution. After the configuration is created, you
train the model (create a solution) by calling the
CreateSolutionVersion operation. Every time you call CreateSolutionVersion
, a new version of the
solution is created.
After creating a solution version, you check its accuracy by calling GetSolutionMetrics. When you are satisfied with the version, you deploy it using CreateCampaign. The campaign provides recommendations to a client through the GetRecommendations API.
To train a model, Amazon Personalize requires training data and a recipe. The
training data comes from the dataset group that you provide in the request. A
recipe specifies the training algorithm and a feature transformation. You can
specify one of the predefined recipes provided by Amazon Personalize.
Alternatively, you can specify performAutoML
and Amazon Personalize will
analyze your data and select the optimum USER_PERSONALIZATION recipe for you.
Amazon Personalize doesn't support configuring the hpoObjective
for solution
hyperparameter optimization at this time.
status
Status
A solution can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
DELETE PENDING > DELETE IN_PROGRESS
To get the status of the solution, call
DescribeSolution. Wait until the status shows as ACTIVE before calling CreateSolutionVersion
.
related-apis
Related APIs
Trains or retrains an active solution in a Custom dataset group.
A solution is created using the
CreateSolution operation and must be in the ACTIVE state before calling
CreateSolutionVersion
. A new version of the solution is created every time you
call this operation.
status
Status
A solution version can be in one of the following states:
CREATE PENDING
CREATE IN_PROGRESS
ACTIVE
CREATE FAILED
CREATE STOPPING
CREATE STOPPED
To get the status of the version, call
DescribeSolutionVersion.
Wait until the status shows as ACTIVE before calling CreateCampaign
.
If the status shows as CREATE FAILED, the response includes a failureReason
key, which describes why the job failed.
related-apis
Related APIs
Removes a campaign by deleting the solution deployment.
The solution that the campaign is based on is not deleted and can be redeployed when needed. A deleted campaign can no longer be specified in a GetRecommendations request. For information on creating campaigns, see CreateCampaign.
Deletes a dataset.
You can't delete a dataset if an associated DatasetImportJob
or
SolutionVersion
is in the CREATE PENDING or IN PROGRESS state. For more
information on datasets, see
CreateDataset.
Deletes a dataset group.
Before you delete a dataset group, you must delete the following:
All associated event trackers.
All associated solutions.
All datasets in the dataset group.
Deletes the event tracker.
Does not delete the event-interactions dataset from the associated dataset group. For more information on event trackers, see CreateEventTracker.
Deletes a filter.
Deletes a metric attribution.
Deactivates and removes a recommender.
A deleted recommender can no longer be specified in a GetRecommendations request.
Deletes a schema.
Before deleting a schema, you must delete all datasets referencing the schema. For more information on schemas, see CreateSchema.
Deletes all versions of a solution and the Solution
object itself.
Before deleting a solution, you must delete all campaigns based on the solution.
To determine what campaigns are using the solution, call
ListCampaigns and supply the Amazon Resource Name (ARN) of the solution. You can't delete a
solution if an associated SolutionVersion
is in the CREATE PENDING or IN
PROGRESS state. For more information on solutions, see
CreateSolution.
Describes the given algorithm.
Gets the properties of a batch inference job including name, Amazon Resource Name (ARN), status, input and output configurations, and the ARN of the solution version used to generate the recommendations.
Gets the properties of a batch segment job including name, Amazon Resource Name (ARN), status, input and output configurations, and the ARN of the solution version used to generate segments.
Describes the given campaign, including its status.
A campaign can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
DELETE PENDING > DELETE IN_PROGRESS
When the status
is CREATE FAILED
, the response includes the failureReason
key, which describes why.
For more information on campaigns, see CreateCampaign.
Describes the given dataset.
For more information on datasets, see CreateDataset.
Describes the dataset export job created by CreateDatasetExportJob, including the export job status.
Describes the given dataset group.
For more information on dataset groups, see CreateDatasetGroup.
Describes the dataset import job created by CreateDatasetImportJob, including the import job status.
Describes an event tracker.
The response includes the trackingId
and status
of the event tracker. For
more information on event trackers, see
CreateEventTracker.
Describes the given feature transformation.
Describes a filter's properties.
Describes a metric attribution.
Describes a recipe.
A recipe contains three items:
An algorithm that trains a model.
Hyperparameters that govern the training.
Feature transformation information for modifying the input data before training.
Amazon Personalize provides a set of predefined recipes. You specify a recipe
when you create a solution with the
CreateSolution API. CreateSolution
trains a model by using the algorithm in the specified
recipe and a training dataset. The solution, when deployed as a campaign, can
provide recommendations using the
GetRecommendations
API.
Describes the given recommender, including its status.
A recommender can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
STOP PENDING > STOP IN_PROGRESS > INACTIVE > START PENDING > START IN_PROGRESS > ACTIVE
DELETE PENDING > DELETE IN_PROGRESS
When the status
is CREATE FAILED
, the response includes the failureReason
key, which describes why.
The modelMetrics
key is null when the recommender is being created or deleted.
For more information on recommenders, see CreateRecommender.
Describes a schema.
For more information on schemas, see CreateSchema.
Describes a solution.
For more information on solutions, see CreateSolution.
Describes a specific version of a solution.
For more information on solutions, see CreateSolution
Gets the metrics for the specified solution version.
Gets a list of the batch inference jobs that have been performed off of a solution version.
Gets a list of the batch segment jobs that have been performed off of a solution version that you specify.
Returns a list of campaigns that use the given solution.
When a solution is not specified, all the campaigns associated with the account are listed. The response provides the properties for each campaign, including the Amazon Resource Name (ARN). For more information on campaigns, see CreateCampaign.
Returns a list of dataset export jobs that use the given dataset.
When a dataset is not specified, all the dataset export jobs associated with the account are listed. The response provides the properties for each dataset export job, including the Amazon Resource Name (ARN). For more information on dataset export jobs, see CreateDatasetExportJob. For more information on datasets, see CreateDataset.
Returns a list of dataset groups.
The response provides the properties for each dataset group, including the Amazon Resource Name (ARN). For more information on dataset groups, see CreateDatasetGroup.
Returns a list of dataset import jobs that use the given dataset.
When a dataset is not specified, all the dataset import jobs associated with the account are listed. The response provides the properties for each dataset import job, including the Amazon Resource Name (ARN). For more information on dataset import jobs, see CreateDatasetImportJob. For more information on datasets, see CreateDataset.
Returns the list of datasets contained in the given dataset group.
The response provides the properties for each dataset, including the Amazon Resource Name (ARN). For more information on datasets, see CreateDataset.
Returns the list of event trackers associated with the account.
The response provides the properties for each event tracker, including the Amazon Resource Name (ARN) and tracking ID. For more information on event trackers, see CreateEventTracker.
Lists all filters that belong to a given dataset group.
Lists the metrics for the metric attribution.
Lists metric attributions.
Returns a list of available recipes.
The response provides the properties for each recipe, including the recipe's Amazon Resource Name (ARN).
Returns a list of recommenders in a given Domain dataset group.
When a Domain dataset group is not specified, all the recommenders associated with the account are listed. The response provides the properties for each recommender, including the Amazon Resource Name (ARN). For more information on recommenders, see CreateRecommender.
Returns the list of schemas associated with the account.
The response provides the properties for each schema, including the Amazon Resource Name (ARN). For more information on schemas, see CreateSchema.
Returns a list of solution versions for the given solution.
When a solution is not specified, all the solution versions associated with the account are listed. The response provides the properties for each solution version, including the Amazon Resource Name (ARN).
Returns a list of solutions that use the given dataset group.
When a dataset group is not specified, all the solutions associated with the account are listed. The response provides the properties for each solution, including the Amazon Resource Name (ARN). For more information on solutions, see CreateSolution.
Get a list of tags attached to a resource.
Starts a recommender that is INACTIVE.
Starting a recommender does not create any new models, but resumes billing and automatic retraining for the recommender.
Stops a recommender that is ACTIVE.
Stopping a recommender halts billing and automatic retraining for the recommender.
Stops creating a solution version that is in a state of CREATE_PENDING or CREATE IN_PROGRESS.
Depending on the current state of the solution version, the solution version state changes as follows:
- CREATE_PENDING > CREATE_STOPPED
or
- CREATE_IN_PROGRESS > CREATE_STOPPING > CREATE_STOPPED
You are billed for all of the training completed up until you stop the solution version creation. You cannot resume creating a solution version once it has been stopped.
Add a list of tags to a resource.
Remove tags that are attached to a resource.
Updates a campaign by either deploying a new solution or changing the value of
the campaign's minProvisionedTPS
parameter.
To update a campaign, the campaign status must be ACTIVE or CREATE FAILED. Check the campaign status using the DescribeCampaign operation.
You can still get recommendations from a campaign while an update is in
progress. The campaign will use the previous solution version and campaign
configuration to generate recommendations until the latest campaign update
status is Active
.
For more information on campaigns, see CreateCampaign.
Updates a metric attribution.
Updates the recommender to modify the recommender configuration.