ルート・スキーマ : schema
型: object
ソースの表示
{
"type":"object",
"required":[
"jobProperties"
],
"properties":{
"jobSchedule":{
"$ref":"#/components/schemas/JobSchedule"
},
"jobProperties":{
"oneOf":[
{
"$ref":"#/components/schemas/DeployedModelScoringJobProperties"
},
{
"$ref":"#/components/schemas/DataMonitoringJobProperties"
},
{
"$ref":"#/components/schemas/ModelMonitoringJobProperties"
}
],
"description":"The properties for this job."
}
}
}
ネストされたスキーマ: jobProperties
このジョブのプロパティ。
1つのスキーマに一致
ソースの表示
{
"oneOf":[
{
"$ref":"#/components/schemas/DeployedModelScoringJobProperties"
},
{
"$ref":"#/components/schemas/DataMonitoringJobProperties"
},
{
"$ref":"#/components/schemas/ModelMonitoringJobProperties"
}
],
"description":"The properties for this job."
}
ネストされたスキーマ: JobSchedule
型: object
ソースの表示
-
jobEndDate: string (date-time)
ジョブの実行を終了する必要がある日時。このプロパティが省略されている場合に、maxRunsプロパティがないと、ジョブは取り消されるまでスケジュールに従って実行されます。ISO-8601日時書式(https://www.iso.org/ISO-8601-date-and-time-format.htmlを参照)にする必要があります
例: 2022-05-12T02:33:16Z
-
jobStartDate: string (date-time)
ジョブの実行を開始する必要がある日時。このプロパティを省略すると、ジョブは即時に開始されます。ISO-8601日時書式(https://www.iso.org/ISO-8601-date-and-time-format.htmlを参照)にする必要があります
例: 2022-05-12T02:33:16Z
-
maxRuns: integer
このスケジュールに従ってジョブを実行できる最大回数。
-
repeatInterval: string
ジョブの実行を繰り返す頻度。この指定は、dbms_schedulerカレンダ構文に従います。たとえば、日次ジョブはFREQ=DAILYと指定します。
{
"type":"object",
"properties":{
"jobStartDate":{
"type":"string",
"format":"date-time",
"description":"The date and time when the job should start execution. If this property is omitted the job will be started immediately. The ISO-8601 date and time format is expected (see https://www.iso.org/iso-8601-date-and-time-format.html)",
"example":"2022-05-12T02:33:16Z"
},
"repeatInterval":{
"type":"string",
"description":"How often the job execution should be repeated. The specification follows the dbms_scheduler calendaring syntax. For example, a daily job would be specified as FREQ=DAILY."
},
"jobEndDate":{
"type":"string",
"format":"date-time",
"description":"The date and time when the job should end execution. If this property is omitted and there is no maxRuns property, then the job will execute on schedule until cancelled. The ISO-8601 date and time format is expected (see https://www.iso.org/iso-8601-date-and-time-format.html)",
"example":"2022-05-12T02:33:16Z"
},
"maxRuns":{
"type":"integer",
"description":"The maximum number of times the job can be executed according to this schedule."
}
}
}
ネストされたスキーマ: DeployedModelScoringJobProperties
型: object
ソースの表示
{
"type":"object",
"required":[
"modelId"
],
"properties":{
"modelId":{
"type":"string",
"description":"The modelId to be used for scoring against."
},
"topNDetails":{
"type":"integer",
"description":"A value for topNDetails.",
"minimum":1,
"maximum":50
}
},
"allOf":[
{
"$ref":"#/components/schemas/ScoringJobDefinitionProperties"
}
]
}
すべて一致
ソースの表示
{
"type":"object",
"required":[
"modelId"
],
"properties":{
"modelId":{
"type":"string",
"description":"The modelId to be used for scoring against."
},
"topNDetails":{
"type":"integer",
"description":"A value for topNDetails.",
"minimum":1,
"maximum":50
}
},
"allOf":[
{
"$ref":"#/components/schemas/ScoringJobDefinitionProperties"
}
]
}
ネストされたスキーマ: DataMonitoringJobProperties
型: object
ソースの表示
-
anchorColumn: string
ベースラインのアンカー列および2変量分析に使用する新規データ。モニター対象の問題のターゲット列を、アンカー列として渡すことができます。
-
featureList: array featureList
データ・モニタリングを実行する特徴のリスト。リスト内の特徴の数は制限する必要があります(デフォルトは250)。特徴リストを指定しないと、すべての特徴でモニタリングが実行されます。250個を超える特徴がある場合は、エラーになります。特徴は数値またはカテゴリ型である必要があります。このリリースでは、テキスト、ネストまたは日付はサポートされていません。
{
"type":"object",
"properties":{
"featureList":{
"type":"array",
"items":{
"type":"string"
},
"description":"A list of features to perform data monitoring on, there should be a limit on the number of features in the list (default 250). Monitoring will run on all features if feature list is not specified. If there are more than 250 features, it will error out. The features need to be either numeric or categorical. Text, nested, or date will not be supported in this release."
},
"anchorColumn":{
"type":"string",
"description":"Anchor column in the baseline and new data used for bi-variate analysis. The target column in supervised problems can be passed as an anchor column."
}
},
"allOf":[
{
"$ref":"#/components/schemas/MonitoringJobDefinitionProperties"
}
]
}
すべて一致
ソースの表示
{
"type":"object",
"properties":{
"featureList":{
"type":"array",
"items":{
"type":"string"
},
"description":"A list of features to perform data monitoring on, there should be a limit on the number of features in the list (default 250). Monitoring will run on all features if feature list is not specified. If there are more than 250 features, it will error out. The features need to be either numeric or categorical. Text, nested, or date will not be supported in this release."
},
"anchorColumn":{
"type":"string",
"description":"Anchor column in the baseline and new data used for bi-variate analysis. The target column in supervised problems can be passed as an anchor column."
}
},
"allOf":[
{
"$ref":"#/components/schemas/MonitoringJobDefinitionProperties"
}
]
}
ネストされたスキーマ: ModelMonitoringJobProperties
型: object
ソースの表示
-
modelList(required): array modelList
モニター対象のモデルのリスト。モデルは、ModelIdで識別されます。このリストの最大許容長は20です。
-
performanceMetric: string
指定できる値: [ "R2", "MEAN_SQUARED_ERROR", "MEAN_ABSOLUTE_ERROR", "MEDIAN_ABSOLUTE_ERROR", "ACCURACY", "BALANCED_ACCURACY", "ROC_AUC", "F1", "PRECISION", "RECALL", "MACRO_F1", "MACRO_PRECISION", "MACRO_RECALL", "WEIGHTED_F1", "WEIGHTED_PRECISION", "WEIGHTED_RECALL" ]
モデル・パフォーマンスの測定に使用されるメトリック。回帰モデルのデフォルトのメトリックはMEAN_SQUARED_ERRORです。分類モデルのデフォルトのメトリックはBALANCED_ACCURACYです。回帰モデルでサポートされているメトリックは、R2、MEAN_SQUARED_ERROR、MEAN_ABSOLUTE_ERRORおよびMEDIAN_ABSOLUTE_ERRORです。ACCURACYおよびBALANCED_ACCURACYは、すべての分類モデルでサポートされています。ROC_AUC、F1、PRECISIONおよびRECALLは、パーティションなしの2項モデルでのみサポートされています。MACRO_F1、MACRO_PRECISION、MACRO_RECALL、WEIGHTED_F1、WEIGHTED_PRECISIONおよびWEIGHTED_RECALLは、多クラス・モデルおよびパーティション化されたモデルでサポートされています。2項分類モデルに対してMACROまたはWEIGHTEDメトリックを選択すると、多クラス・モデルとして扱われます。
{
"type":"object",
"required":[
"modelList"
],
"properties":{
"modelList":{
"type":"array",
"items":{
"type":"string"
},
"description":"A list of models to be monitored. Models are identified by their ModelIds. The maximum length allowed for this list is 20."
},
"performanceMetric":{
"type":"string",
"enum":[
"R2",
"MEAN_SQUARED_ERROR",
"MEAN_ABSOLUTE_ERROR",
"MEDIAN_ABSOLUTE_ERROR",
"ACCURACY",
"BALANCED_ACCURACY",
"ROC_AUC",
"F1",
"PRECISION",
"RECALL",
"MACRO_F1",
"MACRO_PRECISION",
"MACRO_RECALL",
"WEIGHTED_F1",
"WEIGHTED_PRECISION",
"WEIGHTED_RECALL"
],
"description":"The metric used to measure model performance. The default metric for regression models is MEAN_SQUARED_ERROR. The default metric for classification models is BALANCED_ACCURACY. The metrics supported for regression models are R2, MEAN_SQUARED_ERROR, MEAN_ABSOLUTE_ERROR and MEDIAN_ABSOLUTE_ERROR. ACCURACY and BALANCED_ACCURACY are supported for all classification models. ROC_AUC, F1, PRECISION, and RECALL are supported only for binary models without partitions. MACRO_F1, MACRO_PRECISION, MACRO_RECALL, WEIGHTED_F1, WEIGHTED_PRECISION and WEIGHTED_RECALL are supported for multiclass models and for partitioned models. If a MACRO or a WEIGHTED metric is selected for a binary classification model, it will be treated as a multiclass model."
}
},
"allOf":[
{
"$ref":"#/components/schemas/MonitoringJobDefinitionProperties"
}
]
}
すべて一致
ソースの表示
{
"type":"object",
"required":[
"modelList"
],
"properties":{
"modelList":{
"type":"array",
"items":{
"type":"string"
},
"description":"A list of models to be monitored. Models are identified by their ModelIds. The maximum length allowed for this list is 20."
},
"performanceMetric":{
"type":"string",
"enum":[
"R2",
"MEAN_SQUARED_ERROR",
"MEAN_ABSOLUTE_ERROR",
"MEDIAN_ABSOLUTE_ERROR",
"ACCURACY",
"BALANCED_ACCURACY",
"ROC_AUC",
"F1",
"PRECISION",
"RECALL",
"MACRO_F1",
"MACRO_PRECISION",
"MACRO_RECALL",
"WEIGHTED_F1",
"WEIGHTED_PRECISION",
"WEIGHTED_RECALL"
],
"description":"The metric used to measure model performance. The default metric for regression models is MEAN_SQUARED_ERROR. The default metric for classification models is BALANCED_ACCURACY. The metrics supported for regression models are R2, MEAN_SQUARED_ERROR, MEAN_ABSOLUTE_ERROR and MEDIAN_ABSOLUTE_ERROR. ACCURACY and BALANCED_ACCURACY are supported for all classification models. ROC_AUC, F1, PRECISION, and RECALL are supported only for binary models without partitions. MACRO_F1, MACRO_PRECISION, MACRO_RECALL, WEIGHTED_F1, WEIGHTED_PRECISION and WEIGHTED_RECALL are supported for multiclass models and for partitioned models. If a MACRO or a WEIGHTED metric is selected for a binary classification model, it will be treated as a multiclass model."
}
},
"allOf":[
{
"$ref":"#/components/schemas/MonitoringJobDefinitionProperties"
}
]
}
ネストされたスキーマ: ScoringJobDefinitionProperties
型: object
ソースの表示
{
"type":"object",
"required":[
"inputData",
"supplementalColumnNames"
],
"properties":{
"inputData":{
"type":"string",
"description":"The name of the table or view to read input data for the job."
},
"supplementalColumnNames":{
"type":"array",
"items":{
"type":"string",
"description":"An array of columns from the input table/view that can be used to identify rows, or carry information required by the job."
}
},
"topN":{
"type":"integer",
"description":"A value for topN. Default is 1.",
"minimum":1,
"default":1
},
"recompute":{
"type":"boolean",
"default":false,
"description":"A flag whether to replace output table. Default is false."
}
},
"allOf":[
{
"$ref":"#/components/schemas/JobDefinitionProperties"
}
]
}
すべて一致
ソースの表示
{
"type":"object",
"required":[
"inputData",
"supplementalColumnNames"
],
"properties":{
"inputData":{
"type":"string",
"description":"The name of the table or view to read input data for the job."
},
"supplementalColumnNames":{
"type":"array",
"items":{
"type":"string",
"description":"An array of columns from the input table/view that can be used to identify rows, or carry information required by the job."
}
},
"topN":{
"type":"integer",
"description":"A value for topN. Default is 1.",
"minimum":1,
"default":1
},
"recompute":{
"type":"boolean",
"default":false,
"description":"A flag whether to replace output table. Default is false."
}
},
"allOf":[
{
"$ref":"#/components/schemas/JobDefinitionProperties"
}
]
}
ネストされたスキーマ: Discriminator: jobType
型: object
{
"type":"object",
"required":[
"jobType",
"outputData",
"jobName"
],
"properties":{
"jobType":{
"type":"string",
"enum":[
"MODEL_SCORING",
"DATA_MONITORING",
"MODEL_MONITORING"
],
"description":"The type of this property set."
},
"jobServiceLevel":{
"enum":[
"LOW",
"MEDIUM",
"HIGH"
],
"description":"The service level for the job. MEDIUM gives more resources to the job compared to LOW. HIGH gives more resources to the job compared to MEDIUM. The default is LOW."
},
"inputSchemaName":{
"pattern":"^(?:[a-zA-Z0-9_$#]{1,128}|\\\"[^\\\"\\\\0]{1,128}\\\")$",
"type":"string",
"description":"The database schema that owns the input table/view. If not provided, null or an empty string, the input schema will be the same as the username in the request token."
},
"outputSchemaName":{
"pattern":"^(?:[a-zA-Z0-9_$#]{1,128}|\\\"[^\\\"\\\\0]{1,128}\\\")$",
"type":"string",
"description":"The database schema that owns the output table. If not provided, null or an empty string, then the output schema will be the same as the input schema."
},
"outputData":{
"type":"string",
"description":"The output table identifier. The results of the job will be written in an output table named <jobId>_<outputData>."
},
"jobDescription":{
"type":"string",
"description":"A description of the job."
},
"jobName":{
"type":"string",
"description":"A unique name for the job."
},
"disableJob":{
"type":"boolean",
"default":false,
"description":"Whether to disable the job at submission. If not set, the default is false and the job is enabled at submission."
}
},
"discriminator":{
"propertyName":"jobType"
}
}
ソースの表示
{
"type":"object",
"required":[
"jobType",
"outputData",
"jobName"
],
"properties":{
"jobType":{
"type":"string",
"enum":[
"MODEL_SCORING",
"DATA_MONITORING",
"MODEL_MONITORING"
],
"description":"The type of this property set."
},
"jobServiceLevel":{
"enum":[
"LOW",
"MEDIUM",
"HIGH"
],
"description":"The service level for the job. MEDIUM gives more resources to the job compared to LOW. HIGH gives more resources to the job compared to MEDIUM. The default is LOW."
},
"inputSchemaName":{
"pattern":"^(?:[a-zA-Z0-9_$#]{1,128}|\\\"[^\\\"\\\\0]{1,128}\\\")$",
"type":"string",
"description":"The database schema that owns the input table/view. If not provided, null or an empty string, the input schema will be the same as the username in the request token."
},
"outputSchemaName":{
"pattern":"^(?:[a-zA-Z0-9_$#]{1,128}|\\\"[^\\\"\\\\0]{1,128}\\\")$",
"type":"string",
"description":"The database schema that owns the output table. If not provided, null or an empty string, then the output schema will be the same as the input schema."
},
"outputData":{
"type":"string",
"description":"The output table identifier. The results of the job will be written in an output table named <jobId>_<outputData>."
},
"jobDescription":{
"type":"string",
"description":"A description of the job."
},
"jobName":{
"type":"string",
"description":"A unique name for the job."
},
"disableJob":{
"type":"boolean",
"default":false,
"description":"Whether to disable the job at submission. If not set, the default is false and the job is enabled at submission."
}
},
"discriminator":{
"propertyName":"jobType"
}
}
ネストされたスキーマ: featureList
型: array
データ・モニタリングを実行する特徴のリスト。リスト内の特徴の数は制限する必要があります(デフォルトは250)。特徴リストを指定しないと、すべての特徴でモニタリングが実行されます。250個を超える特徴がある場合は、エラーになります。特徴は数値またはカテゴリ型である必要があります。このリリースでは、テキスト、ネストまたは日付はサポートされていません。
ソースの表示
{
"type":"array",
"items":{
"type":"string"
},
"description":"A list of features to perform data monitoring on, there should be a limit on the number of features in the list (default 250). Monitoring will run on all features if feature list is not specified. If there are more than 250 features, it will error out. The features need to be either numeric or categorical. Text, nested, or date will not be supported in this release."
}
ネストされたスキーマ: MonitoringJobDefinitionProperties
型: object
ソースの表示
-
baselineData(required): string
モニター対象のベースライン・データを含む表/ビュー。データ・モニタリングには少なくとも100行が必要です。モデル・モニタリングには少なくとも50行が必要です。行数が少ない場合、分析はスキップされます。
-
caseidColumn: string
ベースラインおよび新しいデータのケース識別子列。指定すると、結果の再現性が向上します。
-
endDate: string (date-time)
newDataのモニタリングの終了日/タイムスタンプ。timeColumnが存在すると仮定しています。endDateを指定しないと、timeColumnの最新の日付が使用されます。ISO-8601日時書式(https://www.iso.org/ISO-8601-date-and-time-format.htmlを参照)にする必要があります
例: 2022-05-12T02:33:16Z
-
frequency: string
新しいデータを分析のためにグループ化する期間を示します。サポートされる値は、"Day"、"Week"、"Month"および"Year"です。分析期間は、endDateから遡って定義されます。startDateで始まる期間には、部分的データが含まれる可能性があります。指定しないと、新しいデータ全体が1つの期間として使用されます。
-
newData(required): string
ベースラインと比較する新しいデータを含む表/ビュー。データ・モニタリングには、期間当たり少なくとも100行が必要です。モデル・モニタリングには、期間当たり少なくとも50行が必要です。行数が少ない場合、期間の分析はスキップされます。
-
recompute: boolean
デフォルト値: false
すでに計算済の期間を更新するかどうかを示すフラグ。デフォルトはfalseです。つまり、結果表のデータよりも新しいデータがある期間だけが計算されます。選択した頻度によっては、新しい分析結果が既存の結果と重複する可能性があります。recomputeがtrueの場合、結果表の指定期間の以前の結果が新しい結果で置き換えられます。
-
startDate: string (date-time)
newDataのモニタリングの開始日/タイムスタンプ。timeColumnが存在すると仮定しています。startDateを指定しない場合は、頻度が指定されているかどうかによってstartDateが異なります。頻度が指定されていない場合は、timeColumnの最も早い日付がstartDateとして使用されます。それ以外の場合は、timeColumnの最も早い日付と最近のサイクルの10番目の開始日のうち、どちらか早い日付がstartDateとして使用されます。ISO-8601日時書式(https://www.iso.org/ISO-8601-date-and-time-format.htmlを参照)にする必要があります。
例: 2022-05-12T02:33:16Z
-
threshold: number (float)
-
timeColumn: string
newData表/ビューに時間情報を格納する列の名前。指定しないと、newData全体が1つの期間として処理されます。サポートされる時間列タイプには、DATE、TIMESTAMP、TIMESTAMP WITH TIME ZONEおよびTIMESTAMP WITH LOCAL TIME ZONEが含まれます。
{
"type":"object",
"required":[
"baselineData",
"newData"
],
"properties":{
"baselineData":{
"type":"string",
"description":"A table/view that contains the baseline data to monitor. At least 100 rows are required for data monitoring. At least 50 rows are required for model monitoring. With fewer rows, the analysis is skipped."
},
"newData":{
"type":"string",
"description":"A table/view with new data to be compared against the baseline. At least 100 rows are required for data monitoring per period. At least 50 rows are required for model monitoring per period. With fewer rows, the analysis for the period is skipped."
},
"timeColumn":{
"type":"string",
"description":"Name of a column storing time info in the newData table/view. If not provided, the entire newData is treated as one period. Supported time column types include DATE, TIMESTAMP, TIMESTAMP WITH TIME ZONE and TIMESTAMP WITH LOCAL TIME ZONE."
},
"startDate":{
"type":"string",
"format":"date-time",
"description":"Start date/timestamp of the monitoring in newData. It assumes the existence of a timeColumn. If startDate is not provided, then the startDate depends on whether frequency has been provided. If frequency has not been provided, the earliest date in the timeColumn is used as the startDate. Otherwise, the more recent of the earliest date in the timeColumn and the starting date of the 10th most recent cycle is used as the startDate. The ISO-8601 date-time format is expected (see https://www.iso.org/iso-8601-date-and-time-format.html).",
"example":"2022-05-12T02:33:16Z"
},
"endDate":{
"type":"string",
"format":"date-time",
"description":"End date/timestamp of the monitoring in newData. It assumes the existence of a timeColumn. If endDate is not provided, the most recent date in the timeColumn will be used. The ISO-8601 date and time format is expected (see https://www.iso.org/iso-8601-date-and-time-format.html).",
"example":"2022-05-12T02:33:16Z"
},
"frequency":{
"type":"string",
"description":"Indicates the length of time during which new data is grouped for analysis. Supported values are \"Day\", \"Week\", \"Month\" and \"Year\". Analysis periods are defined from the endDate going backward. The period starting with startDate may contain partial data. If not provided, the whole new data will be used as a single period."
},
"threshold":{
"type":"number",
"format":"float",
"description":"Threshold to trigger drift alert."
},
"recompute":{
"type":"boolean",
"default":false,
"description":"A flag whether to update already computed periods. Default is false. That is, only periods with newer data than present in the result table will be computed. New analysis results may overlap with the existing results depending on the selected frequency. If recompute is true, then new results will replace previous results for the specified time period in the result table."
},
"caseidColumn":{
"type":"string",
"description":"Case identifier column in the baseline and new data. If provided, it improves the repeatibility of the results."
}
},
"allOf":[
{
"$ref":"#/components/schemas/JobDefinitionProperties"
}
]
}
すべて一致
ソースの表示
{
"type":"object",
"required":[
"baselineData",
"newData"
],
"properties":{
"baselineData":{
"type":"string",
"description":"A table/view that contains the baseline data to monitor. At least 100 rows are required for data monitoring. At least 50 rows are required for model monitoring. With fewer rows, the analysis is skipped."
},
"newData":{
"type":"string",
"description":"A table/view with new data to be compared against the baseline. At least 100 rows are required for data monitoring per period. At least 50 rows are required for model monitoring per period. With fewer rows, the analysis for the period is skipped."
},
"timeColumn":{
"type":"string",
"description":"Name of a column storing time info in the newData table/view. If not provided, the entire newData is treated as one period. Supported time column types include DATE, TIMESTAMP, TIMESTAMP WITH TIME ZONE and TIMESTAMP WITH LOCAL TIME ZONE."
},
"startDate":{
"type":"string",
"format":"date-time",
"description":"Start date/timestamp of the monitoring in newData. It assumes the existence of a timeColumn. If startDate is not provided, then the startDate depends on whether frequency has been provided. If frequency has not been provided, the earliest date in the timeColumn is used as the startDate. Otherwise, the more recent of the earliest date in the timeColumn and the starting date of the 10th most recent cycle is used as the startDate. The ISO-8601 date-time format is expected (see https://www.iso.org/iso-8601-date-and-time-format.html).",
"example":"2022-05-12T02:33:16Z"
},
"endDate":{
"type":"string",
"format":"date-time",
"description":"End date/timestamp of the monitoring in newData. It assumes the existence of a timeColumn. If endDate is not provided, the most recent date in the timeColumn will be used. The ISO-8601 date and time format is expected (see https://www.iso.org/iso-8601-date-and-time-format.html).",
"example":"2022-05-12T02:33:16Z"
},
"frequency":{
"type":"string",
"description":"Indicates the length of time during which new data is grouped for analysis. Supported values are \"Day\", \"Week\", \"Month\" and \"Year\". Analysis periods are defined from the endDate going backward. The period starting with startDate may contain partial data. If not provided, the whole new data will be used as a single period."
},
"threshold":{
"type":"number",
"format":"float",
"description":"Threshold to trigger drift alert."
},
"recompute":{
"type":"boolean",
"default":false,
"description":"A flag whether to update already computed periods. Default is false. That is, only periods with newer data than present in the result table will be computed. New analysis results may overlap with the existing results depending on the selected frequency. If recompute is true, then new results will replace previous results for the specified time period in the result table."
},
"caseidColumn":{
"type":"string",
"description":"Case identifier column in the baseline and new data. If provided, it improves the repeatibility of the results."
}
},
"allOf":[
{
"$ref":"#/components/schemas/JobDefinitionProperties"
}
]
}
ネストされたスキーマ: modelList
型: array
モニター対象のモデルのリスト。モデルは、ModelIdで識別されます。このリストの最大許容長は20です。
ソースの表示
{
"type":"array",
"items":{
"type":"string"
},
"description":"A list of models to be monitored. Models are identified by their ModelIds. The maximum length allowed for this list is 20."
}