In this paper, we propose and critically evaluate a combination of methods coming from the Big Data scientific field in order to face a recurring challenge in managerial research: the identification of start-ups of high growth potential (High Growth Firms). Different machine learning techniques have been applied to analyse a dataset composed by 28,353 start-ups. The results we obtained are partially satisfying and suggest that Big Data methods could fail in the presence of highly unbalanced samples. After re-balancing the sample and using statistical techniques aimed at generating artificial samples, we were able to obtain some promising, even if preliminary, results. Our study advances our knowledge about the use of Big Data science in the managerial research and in understanding the limits and caveats of its use.
Can big data do the job? Using big data analysis to predict the growth potential of start-up firms / Balboni, Bernardo; Bortoluzzi, Guido; Pugliese, Roberto; Kourousias, George. - (2017). (Intervento presentato al convegno R&D Management Conference tenutosi a KU Leuven nel 1-5th of July 2017).
Can big data do the job? Using big data analysis to predict the growth potential of start-up firms
Bernardo Balboni;
2017
Abstract
In this paper, we propose and critically evaluate a combination of methods coming from the Big Data scientific field in order to face a recurring challenge in managerial research: the identification of start-ups of high growth potential (High Growth Firms). Different machine learning techniques have been applied to analyse a dataset composed by 28,353 start-ups. The results we obtained are partially satisfying and suggest that Big Data methods could fail in the presence of highly unbalanced samples. After re-balancing the sample and using statistical techniques aimed at generating artificial samples, we were able to obtain some promising, even if preliminary, results. Our study advances our knowledge about the use of Big Data science in the managerial research and in understanding the limits and caveats of its use.Pubblicazioni consigliate
I metadati presenti in IRIS UNIMORE sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono rilasciati con licenza Attribuzione 4.0 Internazionale (CC BY 4.0), salvo diversa indicazione.
In caso di violazione di copyright, contattare Supporto Iris