Application of subset regression on Marketing Mix Modelling (MMM)
Consulting project for IRI Worldwide, also a dissertation project for MSc. Business Analytics • Built a Marketing Mix Model to assess the impact of marketing campaigns, price and promotions of own and competitive products, effects of holiday and weather on the sales volume of the own product by using the retail audit data • Managed the dataset with 433 variables and handled the full process from data understanding, data cleaning, variable selection, data visualisation, model building and validation to model interpretation. R has been used as the major data analysis tool for the project. • Evaluated different subset regression model selection methods including the common stepwise selection method and exhaustive search by branch-and-bound method. Demonstrated the stepwise selection method cannot select the best subset model because of a model with less error can be found by branch-and-bound method