The Power BI Summit was hosted by RADACAD with some 1000+ attendees and 100+ speakers – a mix of the Power BI Product team, Microsoft MVPs and Professionals.

The sessions were targeted for beginner to intermediate level users with some more advanced topics available. By far the biggest value came from Q&A sessions, hearing from other developers and consultants on their lessons learned for Power BI projects within various organisations.

The Power BI Summit is one of the many events that data analytics professionals around the world look forward to each year. With over 100 sessions offered in this year’s conference, check out which ones were our consultants’ favourites:

Contributors: Jean-Noel Seneque, Emma Girvan and Kerry Kolosko

Power BI Dataflows? Why you need to implement it!

By Jean-Noel

The reason why we have Power BI Dataflows

  • Dataflows provide an abstraction layer that handles all the logic, complexities and query steps
  • Dataflows is a Power Query process that runs in the cloud
  • Self-Service Data Preparation for non-technical user personas

Power BI Dataflows is appealing to power users compared to Power BI Desktop because of the added features that have not been released in the desktop yet.

Visual Query Folding Indicators, essentially the Power Query, send back the transformations to the data source to do the heavy lifting. This helps present a better view of the data to be consumed and less work on the Power BI side.

Visual Query Folding

Diagram View provides an overview of a lineage of the data inside the dataflow. You also get a visual overview of the transformation that is occurring and have the ability to extend the dataflow by adding more data, spreadsheets, flat files to the dataflow. You are not connecting to a live dataset.

Data Refresh?

  • Works the same as Dataset
  • We can trigger it manually or set a schedule
  • It does an upstream stream where it will up the data flows it depends on Incremental refresh is available to premium customers
  • The only issue right now is getting error information, the only place to see this is in the refresh history which is not really intuitive

How does the Query Reusability work?

  • You can copy and paste the query from Power BI Desktop but not Power BI Dataflows
  • Linked/Computed Entities (available to premium/PPU Only). This provides the ability to utilise the In-Lake Compute which helps transform the data in the lake
  • Reference points to the existing entity, uses its data and logic but you can build on top of it
  • You can link to other entities in other dataflows
  • It is always Import mode
  • Security is the responsibility of the dataset designer
  • The only issue right now is getting error information, the only place to see this is in the refresh history which is not really intuitive

Building great data experiences in Power BI – Tips from the Product Team

By Jean-Noel

Dissecting what makes great data experiences and the 8 essential steps to achieve that goal.

Goals of great data experiences

  • Transfer of information, some insights you have found and that the users have interpreted correctly
  • Not just to answer the question being asked but why they asked that question in the first place
  • Good leaders will always keep asking questions and have follow-up questions and a great data experience will help let them answer those questions themselves
  • Think of the decisions being taken as a result of that information being transferred
  • What are the users motivations for making those decisions and how is it connected to the business processes that they are in charge of?

Essential #1 – who are your end-users and what do they need from these reports?

Essential #2 – design your reports with the 4 W’s in mind: Who? Where? When? What? The number one tip is to stay focussed and minimise any clusters and distractions. Check out these layouts and get inspired.

Essentials #3 – Right chart for the right purpose

Essential #4 – Allow report interactivity and exploration

Essential #5 – Translate from insights to actions

Essential #6 – This is a form of communication, make it easy to read and understand

Essential #7 – Make the reports available for the on-the-go users; anywhere, anytime

Essential #8 – Performance and optimisation

Designing for High Performance

By Emma

Here are some common steps to improve data model performance:

  • Avoid using high precision/cardinality columns as it creates a level of detail that impacts the ability for the technology to compress the data effectively (i.e., Date/time fields with seconds/milliseconds, numeric values with many decimal places)
    • Consider separating out the data into separate columns if it is required and use Power Query to join the data back together as required.
    • If you do not need to have an infinite number of decimal places in the source data, create a step to limit the number to only what is required.
  • Consider using composite models instead of pure Direct Query data imports if some of your data does not need to be near real time.
    • A possible solution could be to bring in Dimension table data with import mode and Fact table data with Direct Query mode.
  • Turn off the auto date/time feature in the Global Options and create custom date tables instead. Power BI creates an internal date table behind the scenes for every date/time field in the dataset. This option has a default setting of ON so it is a good idea to turn this off.
  • Disable meaningless interactions between visuals on the page by turning off interactions using the Edit Interactions option.

Bringing Visualisation Grammar to Power BI

By Kerry

This session involved a walk-through of a preview feature visual, which like Charticulator, allows Power BI Report Developers to build their own custom visuals in Power BI.  

  • Unlike Charticulator, the visual is grammar-based, (rather than click and drop) leveraging Vega languages with far more extensive capabilities. 
  • It requires light coding but no programming experience, making it accessible to most. 
  • The visual runs in desktop as you type with no external dependencies like the R and Python extensions. The possibilities for this visual are exciting, as it allows fine grain customisations on interactivity, faceting, and variable parameters. With this visual, developers can plot distributions in ways that are not currently available in native desktop or appsource visualisations, in a fraction of the time it would have taken to develop via other means.  

Combining Multiple Data Sets in Power Query

By Emma

  • Did you know there is a hidden trick (and best practice) for importing multiple worksheets from the same Excel file without clicking each individual worksheet?
  • By clicking on the header name in the import table process and using Power Query steps to ‘clean up’ the dataset, you will be creating a more scalable data import process if more tables need to be added in the future.
  • You will also eliminate duplication of transformations across multiple separate tables.

Power BI Tips, Tricks & Hacks

By Emma

  • Did you know that if you want to list a measure in more than one display folder in the Power BI model, because it may make sense to the report developer to find it in various places, you can do this by separating each display folder name with a semicolon?! It does not duplicate the measure, but simply makes it available for selection from multiple places in the model.
  • Did you know by using calculation groups, you can save a stack of time creating a series of time intelligence measures where the logic is exactly the same, other than the date period? Think of all the many financial measures you have created where you have spent time copying and pasting the same thing and changing 3 little letters for each set.
    • This also allows for the use of dynamically selected measures which can then also have the appropriate field formatting applied if you have a mix of $$ and % values. But use caution as there could be a performance impact.
  • Did you know there is a way to use conditional formatting with a measure on the line chart visual, despite there not being an option in Power BI to do this?
    • Create your visual using a column chart, apply your conditional formatting and then change to a line chart visual.
  • Did you know you can apply transparency colours by adding 2 extra digits to the hex value when creating measure for use in conditional formatting?
    For example:
    Cndf – White Trans = “#FFFFFF00”
    Tip: use 00 for 100% and 80 for 50% (e.g., FFFFFF00, FFFFFF80)

AI in Power BI

By Kerry

This session provided an introduction to AI assisted visuals and delved into ML Capabilities within Power BI Premium. 

  • This was a fantastic session walking through various use cases for the feature set allowing the audience to imagine how these could be applied to their own data.  
  • Anomaly detection, smart narratives and predictive analytic capabilities were also discussed in this session. 
Previous Post Next Post