This study investigates to what extent Web effort estimation models built using cross-company data sets can provide suitable effort estimates for Web projects belonging to another company, when compared to Web effort estimates obtained using that company's own data on their past projects (single-company data set). It extends a previous study (S3) where these same research questions were investigated using data on 67 Web projects from the Tukutuku database. Since S3 was carried out, data on other 128 Web projects was added to Tukutuku; therefore this study uses the entire set of 195 projects from the Tukutuku database, which now also includes new data from other single-company data sets. Predictions between cross-company and single-company models are compared using Manual Stepwise Regression+Linear Regression and Case-Based Reasoning. In addition, we also investigated to what extent applying a filtering mechanism to cross-company datasets prior to building prediction models can affect the accuracy of the effort estimates they provide. The present study corroborates the conclusions of S3 since the cross-company models provided much worse predictions than the single-company models. Moreover, the use of the filtering mechanism significantly improved the prediction accuracy of cross-company models when estimating single-company projects, making it comparable to that using single-company datasets.

Web effort estimation: the value of cross-company data set compared to single-company data set

FERRUCCI, Filomena;SARRO, FEDERICA
2012

Abstract

This study investigates to what extent Web effort estimation models built using cross-company data sets can provide suitable effort estimates for Web projects belonging to another company, when compared to Web effort estimates obtained using that company's own data on their past projects (single-company data set). It extends a previous study (S3) where these same research questions were investigated using data on 67 Web projects from the Tukutuku database. Since S3 was carried out, data on other 128 Web projects was added to Tukutuku; therefore this study uses the entire set of 195 projects from the Tukutuku database, which now also includes new data from other single-company data sets. Predictions between cross-company and single-company models are compared using Manual Stepwise Regression+Linear Regression and Case-Based Reasoning. In addition, we also investigated to what extent applying a filtering mechanism to cross-company datasets prior to building prediction models can affect the accuracy of the effort estimates they provide. The present study corroborates the conclusions of S3 since the cross-company models provided much worse predictions than the single-company models. Moreover, the use of the filtering mechanism significantly improved the prediction accuracy of cross-company models when estimating single-company projects, making it comparable to that using single-company datasets.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11386/3891229
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 18
  • ???jsp.display-item.citation.isi??? ND
social impact