Potential Market Value for Portuguese Territory: Demographics as Mobility Proxy

In retail a business location is of extreme importance as it can directly affect profits. Therefore, it is of the utmost importance to be accurate and critical when allocating products and sales personal to a location. This report is the result of the work performed at PSE during a curricular internship. PSE is a data science company that offers data science services and market research. They are currently developing a model that segments a region according to its Potential Market Value. This model relies on multiple factors, including human mobility data. The challenge lies in implementing the model to new locations for which they do not yet have accurate human mobility data. This was the focus of the work developed, finding a proxy for human mobility by using demographic data pared with information regarding different types of Points of Interest. To tackle the issue at hand data mining techniques were used following the CRISP-DM methodology as well as the Tabu Search Algorithm, culminating in the creation of a Potential Market Segmentation Model that can be reproduced for multiple locations and even other companies and retail industries. The results of the new model exceeded the expectations, and this new approach seems to be quite promising. Although there is a loss of information regarding mobility, there is an increase in the knowledge regarding the consumer profile. It is important to notice that the success of the model is related to the type of product, since the consumption of the product used in this study is highly correlated with the residency area of the consumer which means that loosing information about mobility does not pose a major setback.