I'm curious how the "social media presence" is measured. Depending on granularity of that data, it seems like it would highly correlate with a basic time trend.
That's a very good observation. While developing the model we kept it to the number of followers on The Big Three social media platforms. Based on the number of followers, a composite score was generated and utilized in that format as a variable. I am personally attempting to collect data on interactions, sentiment, impressions, Etc to give this variable for granularity. In terms the model itself and its use historical data the social media presence posed an interesting problem since it wasn't really in use until 2008. Therefore, before using the composite score it was a weaker variable in terms of correlation and effect.
I'm curious how the "social media presence" is measured. Depending on granularity of that data, it seems like it would highly correlate with a basic time trend.
That's a very good observation. While developing the model we kept it to the number of followers on The Big Three social media platforms. Based on the number of followers, a composite score was generated and utilized in that format as a variable. I am personally attempting to collect data on interactions, sentiment, impressions, Etc to give this variable for granularity. In terms the model itself and its use historical data the social media presence posed an interesting problem since it wasn't really in use until 2008. Therefore, before using the composite score it was a weaker variable in terms of correlation and effect.