Datasets are not a new concept in SAP Analytics Cloud, as Smart Predict uses them behind the scenes. Both datasets and models that are based on acquired data will be consumed in the same way in stories. The game-changing difference is in how datasets support you with ad-hoc analysis; datasets are better suited for scenarios when the end-user starts from the data, and not by defining the structure.
Using Datasets for Efficiency and Flexibility
When you upload data in a story, you are using an embedded dataset behind the scenes. With a dataset, there is limited data management of dimensions, no data is deleted during wrangling (data preparation), and your data is stored as a table and separate metadata which frees you to toggle back and forth seamlessly between data preparation and story view in a flexible workflow.
Datasets increase your efficiency by adjusting instantaneously to any change. Datasets get you quickly up and running with the unparalleled power of switching between data preparation view to story view. This supports you with ad-hoc analysis and ability to switch to compelling visualizations within stories, empowering you to react quickly when asked to make changes due to business requests.
More Dataset Details
You can Secure your Dataset
Just like any objects stored in the File Repository, security can be applied at the Dataset object level. For example, if a user opens a story that has been based on a dataset and model, but does not have access to the dataset, then the user will only be able to see the charts based on the model when opening up the story.
You can Save Your Dataset as a Model
Today in SAP Analytics Cloud, a model can be created from a dataset. You can schedule the model against the dataset and refresh the data in the model by keeping it up to date with the dataset.
You can Set Specific User Rights Datasets
By default, every role can view a dataset but can limit creation of a dataset to a specific role. Depending on your use case, your organization may envision deployment strategies where you provide LOBs the ability to create datasets, while the more powerful workflow of model creation is kept in the hands of IT (adding more trust to the models as coming out of a governed process).
A final note on datasets: analytic models based on acquired data and datasets share the same level of functionalities in stories, blending included. However, datasets are not supported in story features using Smart Insight on a variance chart and cross-calculations on tables and charts (just like remote models and universal models).
When Models Are the Way to Go
There are still some cases when you use a model instead of a dataset. For example,
- When connecting to live data, the structure of the data is already defined by the source and inherited, fitting into how models are created.
- In Planning use cases, when the planner already has a structure in mind and decides to input the data or import it from a different source to fit into the model.
- When you need to create models from governed data that IT owns.
Overall, in SAP Analytics Cloud, you can see that datasets and models complement each other: datasets are used for ad-hoc, ungoverned use cases based on acquired data, and models are used for governed cases.
See for yourself the power of using datasets to iteratively work through the full analytical cycle of accessing data, manipulating it, cleansing it, creating analytics, validating different scenarios and sharing your insights with other colleagues in one flexible workflow. Start your SAP Analytics Trial today.