Are you ready to put your organization on a path of continuous improvement using the most valuable decision support techniques? Check out Microsoft Senior Program Manager Christian Berg's timely series of essays on how to become your organization's strategic advisor using Machine Learning and Power BI, available on the Community Blog.
Each post takes on a different aspect of business intelligence, and includes a how-to section to get you started creating customized solutions for your team. Most analysis in the posts is done using the R language, but Christian tries to keep any scripts generic enough that anyone can apply them to their own data even without prior hands-on R experience. The posts also share examples of how businesses use these techniques today to drive deeper insights and better outcomes across a variety of scenarios and industries.
Christian is about half-way through the series, so now is a great time to catch up! We've got links to the first five parts below to get you started.
Part One: Strategic decisions support through business driver analysis
A common problem with relying on standard reports is confirmation bias and existing misunderstandings of true drivers that can persist, or a change in underlying fundamental is overlooked. In this post, Christian begins sharing techniques that he has seen successfully employed to counter common heuristics and to increase the chance of continuous improvement.
Part Two: Facilitating business performance reviews
The complexity of underlying assumptions can make an initial analysis time consuming, difficult to quickly understand, and less informative of the root cause of the variance. A simple analysis that Christian finds very effective, both as a complement and as an input to a large normalization model, was time series decomposition. It is easy to perform and explain, and it illustrates many of the opportunities of Machine Learning. Read this post to get started!
Part Three: Revenue and forecasting
Forecasting is an important part of most management positions. Knowing how many products will sell where and when, informs resource decisions like staffing and inventory. It is also an intuitive way to compare performance between different business units, product categories etc. For P&L owners it is an essential tool to proactively manage their margin by adapting investment levels.
Part Four: Managing complexity
Looking at how various factors relate to one another in the "real world" can be overwhelming. For instance, revenue and inventory may be positively correlated during the high season, but negatively correlated during the other months of the year if a lack of stock hurts sales. A simple correlation plot would not be able to demonstrate these factors. In this post, Christian looks at a simple approach to detect simultaneous interactions between multiple variables.
Part Five: Quantifying business impact of previous investments
Estimating the actual impact of an investment is an essential part of management. An investment, in this context, can be any change in resource allocation but is most often a project or marketing initiative. By quantifying the major benefits realized we obtain an objective measure of the outcome. This is useful when discussing lessons learnt and often invaluable input the next time a similar investment is considered.
Part Six: Visualizing and interacting with your Azure Machine Learning Studio experiments
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. In Part 6, he looked 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. He also looks at a simpler example where we instead use DAX to explore the impact of different discount percentages, based on an assumption about our elasticity.