When I right click on a line chart, it shows Analyze and further shows explain the increase/ decrease. This is excellent.
My idea is that when we further click on explain the increase/ decrease, it should show the analyze should be based on last period or same period last year or any specific period.
Hopeful, the concern of BI team will review my idea and work on it.
Jamal Qamar4 votes
With the new key influencers visual, it's common for it to identify key influencers where the number of records described are only 1 or 2 (for example, "finish type quit is 11x more likely in Hanover, Maryland" when there was only one finish in Hanover Maryland and it was a quit). My idea would be to assign a weight based on number of records and adjust the influencers toward the overall population mean the lower below a threshold (30 records?) the number of records involved is.2 votes
Key Influencers evolution over time.
The new Key Influencers visual does a nice job to understand the contribution of different factors in a given population. If the evaluation is performed over a given period, it is likely that some factors produce a different influence over time, i.e. the mix of factors/influencers tends to be more dynamic than static over time. For example, in a mix of products at different stages of their lifecycle, it is not uncommon to see new products behaving differently to products that initiate their decline. In other cases, some factors have a significant influence triggered by a specific event (e.g. a machine breakdown, a promotion, etc..) and ending sometime after (e.g. breakdown corrected, end of promotion, etc..)
It would be nice to add a time dimension to the Key Influencers analysis to understand how the top key influencers evolve over the evaluated period. One way to visualize the results could be a ‘bump’ chart.
With the new Key Influencers visual, understanding how the different influencers evolve over time requires for example a time slicer that is gradually adjusted while annotating (manually) the key influencers displayed for each selected period. This can become a tedious exercise that could be facilitated by a new or enhanced visual incorporating a time dimension.
Key Influencers evolution over time.
The new Key Influencers visual does a nice job to understand the contribution of different factors in a given population. If the evaluation is performed over a given period, it is likely that some factors produce a different influence over time, i.e. the mix of factors/influencers tends to be more dynamic than static over time. For example, in a mix of products at different stages of their lifecycle, it is not uncommon to see new products behaving differently to products that initiate their decline. In other cases, some factors have a significant influence triggered by…74 votes
Power BI has a variety of easy, built-in ML models/algorithms in desktop, sever, and Q&A. Can we get some simple built in quality check pop-ups? Something like a ROC curve with a Thumbs-up/Thumbs-down symbol or a note that says "Hey, this model fit 100% of your data points. It's probably over-fitting."15 votes
We would like the possibility to add custom R libraries to our Premium capacity for example healthcare.ai 's Machine Learning library.2 votes
Similar to R integration in Power BI, it will so important for C# developers to use the newly emerging open-source .NET ML (Machine Learning) https://github.com/dotnet/machinelearning within Power BI.
Check the Roadmap of this library at https://github.com/dotnet/machinelearning/blob/master/ROADMAP.md
- Binary classification
- Multiclass classification
- Anomaly detection (coming soon)
- Ranking (coming soon)
- Recommendation (coming soon)
Power BI ♥ ML
Naji K.65 votes
We are working on this now and should be available soon.
I want to predict with machine learning model in Power Query.
for quick / low-cost predict without using Azure.
1. Azure Machine Learning can export model as Core ML.
2. T-SQL of SQL Server have predict feature.
3. Model can use on-premises, offline, without using Azure.
So, I hope to importing model to Power Query,
and predict with model in Power Query.
to predict a lot of data on-premises/offline/quick/low-cost.
Program should able to answer What if questions and it should show us a notificaition if the sale is going low than Last year or Last month or Lastquarter or last week, etc as coded.1 vote
Function to "automatically find clusters" allows user to select number of clusters, as option. Can an option to use scalable EM algorithm (default) vs. non-scalable EM vs. K Means be added to the desktop tool?7 votes
Input fields that go into Azure Meachine Learning as WebService Input and gives us a predictive result on PowerBI
Can we have an Input field set on Power BI that goes into the Azure Machine Learning Predictive Experiment, calculates the predictive rice based on the inputs and gives back the results to PowerBI.7 votes
Azure Machine Learning integration for Dataflows is now available for preview: https://docs.microsoft.com/en-us/power-bi/service-machine-learning-integration.
We’re planning to extend this to Power BI desktop.
For machine learning I often have value that is derived from other values but is an "age" for example age_in_hours, where I do not want to sum, average, min, max, but use date-time types. Having a date time measure that starts at a "born on" time of zero and aging would be awesome. This way we could time sequence by days, weeks, months, years as we do with other dates.2 votes
Power BI/Cognitive Services combination could be a killer app but it needs to be made more accessible.
There are third party documented methods for accessing Microsoft Cognitive Services APIs from Power BI but the process is a little convoluted and does not work for asynchronous APIs such as Text Analytics Topic API.
My suggestion would be to add an "Analytics" tab to list the available cognitive services.
Integration of Power BI sign-ins and Cognitive Services sign-ins would also be handy.12 votes
We started the work on this feature, will be released in preview soon
Add common data mining algorithms to Power BI. Solver Algorithms, Linear Regression, Association Rules, and Clustering would be awesome. My understanding is that they were available in the old Power Query and Power Pivot add-ins to Excel.10 votes
1) Click on a "Sherlock Homes" icon (for data scope) and get a bounding rectangle with sub-boxes summarizing interesting patterns and correlations found in the dataset, chart, graph, grid.
2) Drag a line charted against time, into future and see time-series prediction.
3) Click on a "Clustering" symbol for a chart, graph, grid and get the choice for various k-means visualizations.
- Don't see your idea?