Drawing from over 100 data points provided by more than eight million consumers across 40 sectors in 40 countries, we integrate business survey information with diverse behavioral data to generate robust forecasts. The data sets range from current buying/consumption behavior, trends, category performance and future outlook.
One of the methods, dependence, uses probabilistic models where the prediction of outcomes is the central theme. It helps us understand the relationship between multiple predictor variables and the predicted variable, which helps us estimate future outcomes. For example, what would the sales look like over time given the media spends, the reach of sales personnel, catchment, etc., or predict binary outcomes such as the likeliness of customer claims in the insurance sector given their age, health condition, occupation, among others.