Tamanho da fonte:
Analytics and Spatial Statistics: Market Expansion as a Growth Strategy for a Health Products Business in Brazil
Última alteração: 11-10-2017
Resumo
Trilha Acadêmica: Big Data and Applied Retail Analytics (tema CLAV 2017)
This study aimed at exploring which criteria are important to support the expansion strategy of a Brazilian healthcare distribution company focused on Orthopedics. The investigated level of granularity was city (within the whole Brazilian territory), and hence, the idea was to identify what makes a city interesting in terms of potential market. We also intended to answer whether there was a "market score" that could help the company’s executives to choose the best places to expand. To do that, a dataset was built with information gathered from public datasets including health and socioeconomic indicators. The adopted analytical approaches included the use of logistic regression to compute city scores, and a series of regression models including linear (OLS) and the innovative Spatial Auto-regressive models (SAR) and Geographically Weighted Regression (GWR). Cluster analyses (k-means and SKATER algorithms) were also used to find non-disjointed territorial groups of cities, according to a set of variables. Main results confirmed that the use of the spatial analytical approaches showed utility and significantly improvements. The spatial regression models showed increasingly better fit to the data, with the SAR model presenting a lower Akaike information criterion (AIC) (42635.8) and greater R2 (0.85) than the OLS model (42761.3 and 0.84, respectively); and the GWR, a very lower AIC (25748.72) and greater R2 (quasi-global R2 of 0.99) than both the OLS and SAR models. Findings provided useful insight for management, relationship segmentation and prediction that can support the development of a solid expansion strategy.
This study aimed at exploring which criteria are important to support the expansion strategy of a Brazilian healthcare distribution company focused on Orthopedics. The investigated level of granularity was city (within the whole Brazilian territory), and hence, the idea was to identify what makes a city interesting in terms of potential market. We also intended to answer whether there was a "market score" that could help the company’s executives to choose the best places to expand. To do that, a dataset was built with information gathered from public datasets including health and socioeconomic indicators. The adopted analytical approaches included the use of logistic regression to compute city scores, and a series of regression models including linear (OLS) and the innovative Spatial Auto-regressive models (SAR) and Geographically Weighted Regression (GWR). Cluster analyses (k-means and SKATER algorithms) were also used to find non-disjointed territorial groups of cities, according to a set of variables. Main results confirmed that the use of the spatial analytical approaches showed utility and significantly improvements. The spatial regression models showed increasingly better fit to the data, with the SAR model presenting a lower Akaike information criterion (AIC) (42635.8) and greater R2 (0.85) than the OLS model (42761.3 and 0.84, respectively); and the GWR, a very lower AIC (25748.72) and greater R2 (quasi-global R2 of 0.99) than both the OLS and SAR models. Findings provided useful insight for management, relationship segmentation and prediction that can support the development of a solid expansion strategy.
Texto completo:
PDF