What Modeling Method Should Be Used For Media Mix Modeling?
Episode 06
Episode 06 Transcript
Hi, I’m John Colias, Senior Vice President of Decision Analyst’s Advanced Analytics Group, here again with the Media Mix Minute. Today we’re going to talk about the modeling method used in media mix modeling.
To begin with, it’s very important to decompose the time-series variation, the across-time variation from the cross-sectional variation, that would be across brands, across markets, across regions, for example. And the reason it’s important to decompose it is to get an accurate measure of the media activity itself on unit sales. As an example, if you see the bar chart in the slide, you can see that media activity for two brands is increasing over time, represented by the bars increasing in height over time, and on the chart, the graph, we can see that as the media activity increases, we get these bumps in the unit sales, and we have two lines—one for one brand and one for the other. And if we were trying to model this data, and we have this sea of points out there, and we fit a line to the data, we get to dash the line. If we didn’t take into account the variation across brands, the differences of the two brands, then we would get this dashed line, which you can see has a slightly steeper slope than each of the individual brand lines.
And what this means is that if we do this kind of modeling, if we use a modeling method that doesn’t decompose the variation across brands from the variation across time, we would get a bias in the measure of the impact of media activity on the unit sales as represented by the little bit steeper slope that you see in that line. So bottom line is we need to decompose these these two elements: the time series of element and the cross-sectional element.
The way that we do that is to apply a method called mixed modeling. It’s the most common method used to do this in media mix modeling. And by using this media mix, this mixed modeling, we can get a more accurate measure of the impact of media activity on sales.
Thank you.
Presenter
John Colias, Ph.D.
Senior VP Research & Development
As a leader with both university teaching and business consulting experience, John focuses on predictive modeling, prescriptive analytics, and artificial intelligence. As Senior Vice President, Research & Development, at Decision Analyst, John combines academic and business interests to help analytics professionals by offering cutting-edge analytic solutions tempered by business realism. He holds a doctorate in economics from The University of Texas at Austin, with specializations in econometrics and mathematical modeling methods.
