2021/05/18 - Amazon Personalize - 3 updated api methods
Changes Amazon Personalize now supports the ability to optimize a solution for a custom objective in addition to maximizing relevance.
{'solutionConfig': {'optimizationObjective': {'itemAttribute': 'string', 'objectiveSensitivity': 'LOW | ' 'MEDIUM ' '| HIGH ' '| OFF'}}}
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.
Note
Amazon Personalize doesn't support configuring the hpoObjective for solution hyperparameter optimization at this time.
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
ListSolutions
CreateSolutionVersion
DescribeSolution
DeleteSolution
ListSolutionVersions
DescribeSolutionVersion
See also: AWS API Documentation
Request Syntax
client.create_solution( name='string', performHPO=True|False, performAutoML=True|False, recipeArn='string', datasetGroupArn='string', eventType='string', solutionConfig={ 'eventValueThreshold': 'string', 'hpoConfig': { 'hpoObjective': { 'type': 'string', 'metricName': 'string', 'metricRegex': 'string' }, 'hpoResourceConfig': { 'maxNumberOfTrainingJobs': 'string', 'maxParallelTrainingJobs': 'string' }, 'algorithmHyperParameterRanges': { 'integerHyperParameterRanges': [ { 'name': 'string', 'minValue': 123, 'maxValue': 123 }, ], 'continuousHyperParameterRanges': [ { 'name': 'string', 'minValue': 123.0, 'maxValue': 123.0 }, ], 'categoricalHyperParameterRanges': [ { 'name': 'string', 'values': [ 'string', ] }, ] } }, 'algorithmHyperParameters': { 'string': 'string' }, 'featureTransformationParameters': { 'string': 'string' }, 'autoMLConfig': { 'metricName': 'string', 'recipeList': [ 'string', ] }, 'optimizationObjective': { 'itemAttribute': 'string', 'objectiveSensitivity': 'LOW'|'MEDIUM'|'HIGH'|'OFF' } } )
string
[REQUIRED]
The name for the solution.
boolean
Whether to perform hyperparameter optimization (HPO) on the specified or selected recipe. The default is false .
When performing AutoML, this parameter is always true and you should not set it to false .
boolean
Whether to perform automated machine learning (AutoML). The default is false . For this case, you must specify recipeArn .
When set to true , Amazon Personalize analyzes your training data and selects the optimal USER_PERSONALIZATION recipe and hyperparameters. In this case, you must omit recipeArn . Amazon Personalize determines the optimal recipe by running tests with different values for the hyperparameters. AutoML lengthens the training process as compared to selecting a specific recipe.
string
The ARN of the recipe to use for model training. Only specified when performAutoML is false.
string
[REQUIRED]
The Amazon Resource Name (ARN) of the dataset group that provides the training data.
string
When your have multiple event types (using an EVENT_TYPE schema field), this parameter specifies which event type (for example, 'click' or 'like') is used for training the model.
If you do not provide an eventType , Amazon Personalize will use all interactions for training with equal weight regardless of type.
dict
The configuration to use with the solution. When performAutoML is set to true, Amazon Personalize only evaluates the autoMLConfig section of the solution configuration.
Note
Amazon Personalize doesn't support configuring the hpoObjective at this time.
eventValueThreshold (string) --
Only events with a value greater than or equal to this threshold are used for training a model.
hpoConfig (dict) --
Describes the properties for hyperparameter optimization (HPO).
hpoObjective (dict) --
The metric to optimize during HPO.
Note
Amazon Personalize doesn't support configuring the hpoObjective at this time.
type (string) --
The type of the metric. Valid values are Maximize and Minimize .
metricName (string) --
The name of the metric.
metricRegex (string) --
A regular expression for finding the metric in the training job logs.
hpoResourceConfig (dict) --
Describes the resource configuration for HPO.
maxNumberOfTrainingJobs (string) --
The maximum number of training jobs when you create a solution version. The maximum value for maxNumberOfTrainingJobs is 40 .
maxParallelTrainingJobs (string) --
The maximum number of parallel training jobs when you create a solution version. The maximum value for maxParallelTrainingJobs is 10 .
algorithmHyperParameterRanges (dict) --
The hyperparameters and their allowable ranges.
integerHyperParameterRanges (list) --
The integer-valued hyperparameters and their ranges.
(dict) --
Provides the name and range of an integer-valued hyperparameter.
name (string) --
The name of the hyperparameter.
minValue (integer) --
The minimum allowable value for the hyperparameter.
maxValue (integer) --
The maximum allowable value for the hyperparameter.
continuousHyperParameterRanges (list) --
The continuous hyperparameters and their ranges.
(dict) --
Provides the name and range of a continuous hyperparameter.
name (string) --
The name of the hyperparameter.
minValue (float) --
The minimum allowable value for the hyperparameter.
maxValue (float) --
The maximum allowable value for the hyperparameter.
categoricalHyperParameterRanges (list) --
The categorical hyperparameters and their ranges.
(dict) --
Provides the name and range of a categorical hyperparameter.
name (string) --
The name of the hyperparameter.
values (list) --
A list of the categories for the hyperparameter.
(string) --
algorithmHyperParameters (dict) --
Lists the hyperparameter names and ranges.
(string) --
(string) --
featureTransformationParameters (dict) --
Lists the feature transformation parameters.
(string) --
(string) --
autoMLConfig (dict) --
The AutoMLConfig object containing a list of recipes to search when AutoML is performed.
metricName (string) --
The metric to optimize.
recipeList (list) --
The list of candidate recipes.
(string) --
optimizationObjective (dict) --
Describes the additional objective for the solution, such as maximizing streaming minutes or increasing revenue. For more information see Optimizing a solution.
itemAttribute (string) --
The numerical metadata column in an Items dataset related to the optimization objective. For example, VIDEO_LENGTH (to maximize streaming minutes), or PRICE (to maximize revenue).
objectiveSensitivity (string) --
Specifies how Amazon Personalize balances the importance of your optimization objective versus relevance.
dict
Response Syntax
{ 'solutionArn': 'string' }
Response Structure
(dict) --
solutionArn (string) --
The ARN of the solution.
{'solution': {'solutionConfig': {'optimizationObjective': {'itemAttribute': 'string', 'objectiveSensitivity': 'LOW ' '| ' 'MEDIUM ' '| ' 'HIGH ' '| ' 'OFF'}}}}
Describes a solution. For more information on solutions, see CreateSolution.
See also: AWS API Documentation
Request Syntax
client.describe_solution( solutionArn='string' )
string
[REQUIRED]
The Amazon Resource Name (ARN) of the solution to describe.
dict
Response Syntax
{ 'solution': { 'name': 'string', 'solutionArn': 'string', 'performHPO': True|False, 'performAutoML': True|False, 'recipeArn': 'string', 'datasetGroupArn': 'string', 'eventType': 'string', 'solutionConfig': { 'eventValueThreshold': 'string', 'hpoConfig': { 'hpoObjective': { 'type': 'string', 'metricName': 'string', 'metricRegex': 'string' }, 'hpoResourceConfig': { 'maxNumberOfTrainingJobs': 'string', 'maxParallelTrainingJobs': 'string' }, 'algorithmHyperParameterRanges': { 'integerHyperParameterRanges': [ { 'name': 'string', 'minValue': 123, 'maxValue': 123 }, ], 'continuousHyperParameterRanges': [ { 'name': 'string', 'minValue': 123.0, 'maxValue': 123.0 }, ], 'categoricalHyperParameterRanges': [ { 'name': 'string', 'values': [ 'string', ] }, ] } }, 'algorithmHyperParameters': { 'string': 'string' }, 'featureTransformationParameters': { 'string': 'string' }, 'autoMLConfig': { 'metricName': 'string', 'recipeList': [ 'string', ] }, 'optimizationObjective': { 'itemAttribute': 'string', 'objectiveSensitivity': 'LOW'|'MEDIUM'|'HIGH'|'OFF' } }, 'autoMLResult': { 'bestRecipeArn': 'string' }, 'status': 'string', 'creationDateTime': datetime(2015, 1, 1), 'lastUpdatedDateTime': datetime(2015, 1, 1), 'latestSolutionVersion': { 'solutionVersionArn': 'string', 'status': 'string', 'creationDateTime': datetime(2015, 1, 1), 'lastUpdatedDateTime': datetime(2015, 1, 1), 'failureReason': 'string' } } }
Response Structure
(dict) --
solution (dict) --
An object that describes the solution.
name (string) --
The name of the solution.
solutionArn (string) --
The ARN of the solution.
performHPO (boolean) --
Whether to perform hyperparameter optimization (HPO) on the chosen recipe. The default is false .
performAutoML (boolean) --
When true, Amazon Personalize performs a search for the best USER_PERSONALIZATION recipe from the list specified in the solution configuration ( recipeArn must not be specified). When false (the default), Amazon Personalize uses recipeArn for training.
recipeArn (string) --
The ARN of the recipe used to create the solution.
datasetGroupArn (string) --
The Amazon Resource Name (ARN) of the dataset group that provides the training data.
eventType (string) --
The event type (for example, 'click' or 'like') that is used for training the model. If no eventType is provided, Amazon Personalize uses all interactions for training with equal weight regardless of type.
solutionConfig (dict) --
Describes the configuration properties for the solution.
eventValueThreshold (string) --
Only events with a value greater than or equal to this threshold are used for training a model.
hpoConfig (dict) --
Describes the properties for hyperparameter optimization (HPO).
hpoObjective (dict) --
The metric to optimize during HPO.
Note
Amazon Personalize doesn't support configuring the hpoObjective at this time.
type (string) --
The type of the metric. Valid values are Maximize and Minimize .
metricName (string) --
The name of the metric.
metricRegex (string) --
A regular expression for finding the metric in the training job logs.
hpoResourceConfig (dict) --
Describes the resource configuration for HPO.
maxNumberOfTrainingJobs (string) --
The maximum number of training jobs when you create a solution version. The maximum value for maxNumberOfTrainingJobs is 40 .
maxParallelTrainingJobs (string) --
The maximum number of parallel training jobs when you create a solution version. The maximum value for maxParallelTrainingJobs is 10 .
algorithmHyperParameterRanges (dict) --
The hyperparameters and their allowable ranges.
integerHyperParameterRanges (list) --
The integer-valued hyperparameters and their ranges.
(dict) --
Provides the name and range of an integer-valued hyperparameter.
name (string) --
The name of the hyperparameter.
minValue (integer) --
The minimum allowable value for the hyperparameter.
maxValue (integer) --
The maximum allowable value for the hyperparameter.
continuousHyperParameterRanges (list) --
The continuous hyperparameters and their ranges.
(dict) --
Provides the name and range of a continuous hyperparameter.
name (string) --
The name of the hyperparameter.
minValue (float) --
The minimum allowable value for the hyperparameter.
maxValue (float) --
The maximum allowable value for the hyperparameter.
categoricalHyperParameterRanges (list) --
The categorical hyperparameters and their ranges.
(dict) --
Provides the name and range of a categorical hyperparameter.
name (string) --
The name of the hyperparameter.
values (list) --
A list of the categories for the hyperparameter.
(string) --
algorithmHyperParameters (dict) --
Lists the hyperparameter names and ranges.
(string) --
(string) --
featureTransformationParameters (dict) --
Lists the feature transformation parameters.
(string) --
(string) --
autoMLConfig (dict) --
The AutoMLConfig object containing a list of recipes to search when AutoML is performed.
metricName (string) --
The metric to optimize.
recipeList (list) --
The list of candidate recipes.
(string) --
optimizationObjective (dict) --
Describes the additional objective for the solution, such as maximizing streaming minutes or increasing revenue. For more information see Optimizing a solution.
itemAttribute (string) --
The numerical metadata column in an Items dataset related to the optimization objective. For example, VIDEO_LENGTH (to maximize streaming minutes), or PRICE (to maximize revenue).
objectiveSensitivity (string) --
Specifies how Amazon Personalize balances the importance of your optimization objective versus relevance.
autoMLResult (dict) --
When performAutoML is true, specifies the best recipe found.
bestRecipeArn (string) --
The Amazon Resource Name (ARN) of the best recipe.
status (string) --
The status of the solution.
A solution can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
DELETE PENDING > DELETE IN_PROGRESS
creationDateTime (datetime) --
The creation date and time (in Unix time) of the solution.
lastUpdatedDateTime (datetime) --
The date and time (in Unix time) that the solution was last updated.
latestSolutionVersion (dict) --
Describes the latest version of the solution, including the status and the ARN.
solutionVersionArn (string) --
The Amazon Resource Name (ARN) of the solution version.
status (string) --
The status of the solution version.
A solution version can be in one of the following states:
CREATE PENDING > CREATE IN_PROGRESS > ACTIVE -or- CREATE FAILED
creationDateTime (datetime) --
The date and time (in Unix time) that this version of a solution was created.
lastUpdatedDateTime (datetime) --
The date and time (in Unix time) that the solution version was last updated.
failureReason (string) --
If a solution version fails, the reason behind the failure.
{'solutionVersion': {'solutionConfig': {'optimizationObjective': {'itemAttribute': 'string', 'objectiveSensitivity': 'LOW ' '| ' 'MEDIUM ' '| ' 'HIGH ' '| ' 'OFF'}}}}
Describes a specific version of a solution. For more information on solutions, see CreateSolution.
See also: AWS API Documentation
Request Syntax
client.describe_solution_version( solutionVersionArn='string' )
string
[REQUIRED]
The Amazon Resource Name (ARN) of the solution version.
dict
Response Syntax
{ 'solutionVersion': { 'solutionVersionArn': 'string', 'solutionArn': 'string', 'performHPO': True|False, 'performAutoML': True|False, 'recipeArn': 'string', 'eventType': 'string', 'datasetGroupArn': 'string', 'solutionConfig': { 'eventValueThreshold': 'string', 'hpoConfig': { 'hpoObjective': { 'type': 'string', 'metricName': 'string', 'metricRegex': 'string' }, 'hpoResourceConfig': { 'maxNumberOfTrainingJobs': 'string', 'maxParallelTrainingJobs': 'string' }, 'algorithmHyperParameterRanges': { 'integerHyperParameterRanges': [ { 'name': 'string', 'minValue': 123, 'maxValue': 123 }, ], 'continuousHyperParameterRanges': [ { 'name': 'string', 'minValue': 123.0, 'maxValue': 123.0 }, ], 'categoricalHyperParameterRanges': [ { 'name': 'string', 'values': [ 'string', ] }, ] } }, 'algorithmHyperParameters': { 'string': 'string' }, 'featureTransformationParameters': { 'string': 'string' }, 'autoMLConfig': { 'metricName': 'string', 'recipeList': [ 'string', ] }, 'optimizationObjective': { 'itemAttribute': 'string', 'objectiveSensitivity': 'LOW'|'MEDIUM'|'HIGH'|'OFF' } }, 'trainingHours': 123.0, 'trainingMode': 'FULL'|'UPDATE', 'tunedHPOParams': { 'algorithmHyperParameters': { 'string': 'string' } }, 'status': 'string', 'failureReason': 'string', 'creationDateTime': datetime(2015, 1, 1), 'lastUpdatedDateTime': datetime(2015, 1, 1) } }
Response Structure
(dict) --
solutionVersion (dict) --
The solution version.
solutionVersionArn (string) --
The ARN of the solution version.
solutionArn (string) --
The ARN of the solution.
performHPO (boolean) --
Whether to perform hyperparameter optimization (HPO) on the chosen recipe. The default is false .
performAutoML (boolean) --
When true, Amazon Personalize searches for the most optimal recipe according to the solution configuration. When false (the default), Amazon Personalize uses recipeArn .
recipeArn (string) --
The ARN of the recipe used in the solution.
eventType (string) --
The event type (for example, 'click' or 'like') that is used for training the model.
datasetGroupArn (string) --
The Amazon Resource Name (ARN) of the dataset group providing the training data.
solutionConfig (dict) --
Describes the configuration properties for the solution.
eventValueThreshold (string) --
Only events with a value greater than or equal to this threshold are used for training a model.
hpoConfig (dict) --
Describes the properties for hyperparameter optimization (HPO).
hpoObjective (dict) --
The metric to optimize during HPO.
Note
Amazon Personalize doesn't support configuring the hpoObjective at this time.
type (string) --
The type of the metric. Valid values are Maximize and Minimize .
metricName (string) --
The name of the metric.
metricRegex (string) --
A regular expression for finding the metric in the training job logs.
hpoResourceConfig (dict) --
Describes the resource configuration for HPO.
maxNumberOfTrainingJobs (string) --
The maximum number of training jobs when you create a solution version. The maximum value for maxNumberOfTrainingJobs is 40 .
maxParallelTrainingJobs (string) --
The maximum number of parallel training jobs when you create a solution version. The maximum value for maxParallelTrainingJobs is 10 .
algorithmHyperParameterRanges (dict) --
The hyperparameters and their allowable ranges.
integerHyperParameterRanges (list) --
The integer-valued hyperparameters and their ranges.
(dict) --
Provides the name and range of an integer-valued hyperparameter.
name (string) --
The name of the hyperparameter.
minValue (integer) --
The minimum allowable value for the hyperparameter.
maxValue (integer) --
The maximum allowable value for the hyperparameter.
continuousHyperParameterRanges (list) --
The continuous hyperparameters and their ranges.
(dict) --
Provides the name and range of a continuous hyperparameter.
name (string) --
The name of the hyperparameter.
minValue (float) --
The minimum allowable value for the hyperparameter.
maxValue (float) --
The maximum allowable value for the hyperparameter.
categoricalHyperParameterRanges (list) --
The categorical hyperparameters and their ranges.
(dict) --
Provides the name and range of a categorical hyperparameter.
name (string) --
The name of the hyperparameter.
values (list) --
A list of the categories for the hyperparameter.
(string) --
algorithmHyperParameters (dict) --
Lists the hyperparameter names and ranges.
(string) --
(string) --
featureTransformationParameters (dict) --
Lists the feature transformation parameters.
(string) --
(string) --
autoMLConfig (dict) --
The AutoMLConfig object containing a list of recipes to search when AutoML is performed.
metricName (string) --
The metric to optimize.
recipeList (list) --
The list of candidate recipes.
(string) --
optimizationObjective (dict) --
Describes the additional objective for the solution, such as maximizing streaming minutes or increasing revenue. For more information see Optimizing a solution.
itemAttribute (string) --
The numerical metadata column in an Items dataset related to the optimization objective. For example, VIDEO_LENGTH (to maximize streaming minutes), or PRICE (to maximize revenue).
objectiveSensitivity (string) --
Specifies how Amazon Personalize balances the importance of your optimization objective versus relevance.
trainingHours (float) --
The time used to train the model. You are billed for the time it takes to train a model. This field is visible only after Amazon Personalize successfully trains a model.
trainingMode (string) --
The scope of training to be performed when creating the solution version. The FULL option trains the solution version based on the entirety of the input solution's training data, while the UPDATE option processes only the data that has changed in comparison to the input solution. Choose UPDATE when you want to incrementally update your solution version instead of creating an entirely new one.
Warning
The UPDATE option can only be used when you already have an active solution version created from the input solution using the FULL option and the input solution was trained with the User-Personalization recipe or the HRNN-Coldstart recipe.
tunedHPOParams (dict) --
If hyperparameter optimization was performed, contains the hyperparameter values of the best performing model.
algorithmHyperParameters (dict) --
A list of the hyperparameter values of the best performing model.
(string) --
(string) --
status (string) --
The status of the solution version.
A solution version can be in one of the following states:
CREATE PENDING
CREATE IN_PROGRESS
ACTIVE
CREATE FAILED
failureReason (string) --
If training a solution version fails, the reason for the failure.
creationDateTime (datetime) --
The date and time (in Unix time) that this version of the solution was created.
lastUpdatedDateTime (datetime) --
The date and time (in Unix time) that the solution was last updated.