Microsoft Power BI Key Influencers for Housing Data

By - February 21, 2020

Microsoft Power BI Key Influencers for Housing Data

Hi, I’m George Casey and I’m at the RSM Technology Experience Center in Denver talking to you today about the key influencers feature and Microsoft Power BI. Now this feature leverages advanced machine learning techniques like multivariate linear regression and clustering algorithms to help us understand and surface these out of the box intelligent analysis so that we can gain insight that we can convert to action.

Now in this case we’re looking at a regression analysis where the dependent variable is the house sales price with several independent variables that might help us predict that such as kitchen quality, garage condition, or lot area. Now with this control in our Power BI dashboard, I can understand how each of these variables can predict or in this case increase the average sales price. So we see kitchen quality, for example, is the highest.

When it’s excellent, we see the average of sales price increases by $159000. Now if I were to drill in a little on that, I can understand how does that vary by kitchen condition? So you see here, excellent is $325000 average. Whereas good is just over $200000. We see that big ramp off, which we can understand intuitively, but here the data provides us exactly how much of a difference that makes.

The other benefit we can take advantage of with this capability is how do we understand this in the context of different segments of selling price? So here we see the system has created these clusters of five segments, each between we’ll say 70 and 110 houses, with different average selling prices. So if I were to go into segment one here, I first understand what are the characteristics that define the segment? In this case, this segment’s defined where the kitchen quality is excellent and the overall quality is greater than a seven on their scale. We see this segment then has a much higher 180000 units higher than the average selling price. And we see how important are kind of that kitchen quality to make up part of that segment.

And conversely, if I drill down to the small segment, segment five, we see this is defined when we have a lot area between 8600-10000 feet, overall quality is less than or equal to seven, and overall quality is greater than six. So we create that small range. And now we see a much closer to what our average overall segment. So the key in this kind of outcome and analysis is that I can very quickly in a drag and drop format, create these quick insights, understand different influencers and experience as I kind of interact with this as opposed to seeing this just in a spreadsheet. And I think this is a very powerful way to take advantage of some of these tried and true techniques like clustering and regression.

Ready to take the next step? Contact us. You may also contact us by telephone: 800.274.3978.

Collaborative leader, data scientist, and problem solver aligning clients with technology and process. Specialties include predictive analytics, marketing automation, CRM, and ERP.

Receive Posts by Email

Subscribe and receive notifications of new posts by email.