The timeseries component is pre-calculated from part two.įor this report to work in Power BI Desktop you need the packages AzureML and ggplot2 installed. We will also look at a simpler example where we instead use DAX to explore the impact of different discount percentages, based on an assumption about our elasticity.įor the example report we have an ensemble regression model that was built in ML Studio to predict revenue by store, product and month using historic data, marketing information and future prices. In this example we will look at connecting to an Azure ML Studio experiment with an Rviz and then building on that to create a dynamic report to explore cross price elasticities. This enables powerful What-If Analysis scenarios where you can use R and ML Studio to assess the distribution of likely outcomes based on the counterfactuals. Using an Rviz has the advantage that you can dynamically select the subset of the Power BI model that you want to score. With Power BI Desktop this can be done in two ways, either by writing the results from the model to a database that you then connect Power BI to, or by using an R script visual (Rviz). Visualizing and exploring the results of your Azure Machine Learning Studio (ML Studio) experiments is useful both when developing and evaluating the model but most importantly when deploying your model and presenting the results. If you need to catch up, you can find the five previous chapters summarized here on the blog.
Microsoft Senior Program Manager Christian Berg is back with another entry in his series on becoming your organization’s strategic advisor with Machine Learning and Power BI.
0 kommentar(er)
