Thing
|
Service
|
Change
|
Reason
|
||
---|---|---|---|---|---|
Data
Training
Signals
Profiles
Prediction
Prescriptive
Prediction
Prescriptive
|
CreateDataset
CreateJob
CreateJob
CreateJob
RealtimeScore
RealtimeScore
BatchScore
BatchScore
|
When using a file from a file system or from a ThingWorx file repository, post bodies for all service request endpoints support only the use of relative filepaths for CSV and JSON locations.
See the CreateDataset Request Samples below.
|
To protect file operating system information by not disclosing absolute file paths in the body of requests.
|
||
Training
|
CreateJob
|
Because the PMML specifications stipulate in particular that the character “.” represents a decimal in the presentation of real number values,ThingWorx Analyticswill enforce English locale settings for PMML generation.
Existing models are migrated automatically during Analytics Server upgrade.
|
To correct inaccuracies in results that stem from using number formats that are not consistent with PMML standards.
|
||
Training
Validation
|
CreateJob
|
Update the confidence level parameter to plural: confidenceLevel -> confidenceLevels
|
To allow creation of multiple confidence models in a single job.
|
||
Validation
|
RetrieveResult
|
Flip the presentation of Boolean confusion matrix results in Analytics Server so that predicted values are shown in rows and actual values in columns.
Existing confusion matrix tables are migrated automatically during Analytics Server upgrade.
See the Confusion Matrix Boolean Samples below.
|
To match the way confusion matrix results are displayed for Categorical and Ordinal opTypes.
|
||
Signals
|
RetrieveResult
|
Add the redundantWithFeature column to the results infotable.
Rename the following columns:
• goal field -> goal
• fields -> features
|
To align with the results retrieved via the SaveJobResult service in the Signals Thing, which outputs a JSON file to the AnalyticsResultsStorage repository.
|
||
Prediction
Prescriptive
|
RealtimeScore
|
Discontinue use of the following Data Shapes to output real time scoring results:
• prediction: AnalyticsPredictionScores
• prescriptive: AnalyticsOptimizationResult
These Data Shapes included nested Infotables that could be adapted to any set of key/value pairs. The new Infotable is model- and database-specific. It generates realtime scoring results in a simpler format but requires an additional procedure to parse the results for consumption by subsequent processes.
See the RealtimeScore Results Samples below.
|
To align the output format of real time scoring results retrieved via the RealtimeScore service with the batch scoring results retrieved via the RetrieveResult service.
|
Old API Request
|
New API Request
|
---|---|
Old Boolean Confusion Matrix
|
New Boolean Confusion Matrix
|
---|---|
Old RealTime Scoring Infotable
|
New RealTime Scoring Infotable
|
---|---|