Predictive Learning ensures a certain level of genericity is maintained in model execution. Only sources with previously defined input and output parameters can be added, and the following source types are supported:
- Predictive Learning Storage
- Data Lake
Adding a New Source¶
To add a new data source, click the + Add New button as shown in the following example:
About Scheduling an IoT Data Export¶
If your model execution requires periodic exports and access to IoT data, the New Source window provides the fields for defining the export. The data export executes at the time you specify and you can also define the properties, and the period of time to go back in history, starting from the export execution time. When the schedule is run, the system writes the exported data to the instance that caries the job execution.
About Integrated Data Lake (IDL) Source Paths¶
For exports that require direct access to a specific path in the data lake, the New Source window provides the fields for defining that path, but the path must already exist in the data lake. If the path doesn't already exist, you must log in to the Integrated Data Lake and create it; there's no way to create the path from outside IDL. When the IDL path is used as an output parameter in scheduling a job, the system uploads the resulting data files to the path you set in the New Source dialog box.
About Predictive Learning Storage Sources¶
Predictive Learning storage (PrL storage) functions similarly to IDL, except that, with PrL storage, you can define the storage path from within the New Source pop-up window, and the system copies the output data to the path you define.
PrL storage is available for all jobs, so there is no need to define it specifically as an input source, as it is always active. Users are free to copy data from PrL storage as needed.
Any questions left?
Except where otherwise noted, content on this site is licensed under the MindSphere Development License Agreement.