Building predictive models isn't about mathematics, or about fancy simulations with hundreds of variables. We believe, the best predictive models are the ones where we understand and clearly identify the factor (f) that is responsible for driving an ongoing shift in the mindset of the consumer.
To learn more about our predictive capabilities, read our white paper on forecasting.
As a political party in the United States prepares for the primaries, they want to know how best they can build a model that will help them better predict the outcome of the primaries and the general election.
Our team conducts a digital ethnographic analysis of over 8500 voters across the United States, and identifies "fairness" and "corporate collusion" as the critical factors affecting voters' choices. In particular, our team identifies swing and undecided voters as being most impacted by these issues.
Our team then identifies a series of 12 variables that are most impacted by the factors identified earlier. These variables serve as inputs into our predictive model. Variables such as mentions/usage of fairness related campaign promises, resulting turnout at rallies, impact on the social media volume of conversation, individual campaign donations received when fairness-related promises are made etc., are all factored into the model to understand why a candidate would be recognized by voters as being fair and able to reduce collusion and the role of special interests in government.
We identify a series of scenarios, if-then-statements, that predict possible outcomes based on the chosen variables. These scenarios are then updated as new information and new campaign promises impact these variables. The net result is that just prior to the end of the primary season, our team accurately identifies the winner of the general election, despite every other mathematical model indicating otherwise.