Output of the fitting process in C4ISTAR domain

Published: 25 June 2020| Version 3 | DOI: 10.17632/yhwhn8zbm5.3
Contributors:
Vito Giordano,
Irene Spada,
Filippo Chiarello,
Gualtiero Fantoni,
Andrea Bonaccorsi

Description

This dataset contains the C4ISTAR technologies that have more than 15 years old extracted with text mining techniques. For each technology we attempt to forecast the potential direction using the S-shaped model. The patent information is used for fitting the S-curves (or growth curves), in particular the Pearl (or Logistic) and Gompertz models. For each technology we indicate: - the error metrics of the growth models used to forecast the future trend of the C4ISTAR technology. For each technology is reported the RMSE and MAE for Logistic (Pearl) model and Gompertz in scientific notation. The errors of the models are reported in "Error" sheet. - the estimated parameters for the best models considering the minimum value of RMSE between the Logistic and Gompertz models. For each parameter we show the standard error and the p-value of the parameters calculated in training model process. The parameters of the models are reported in "Parameter" sheet. The model parameters for either Logistic and Gompertz curves are: - d the upper limit (or asymptote) of the cumulative patent numbers; - b the parameter that depend on the growth in the inflection points; - a the year of inflection. In the "Parameter" sheet we also indicate for each technology the year of birth, the year of inflection calculated as the parameter a plus the year of birth, the actual value of the overall number of patents at year 2017 and the maximum potential value of the cumulative number of patents, equal to the parameter d rounded.

Files