Algorithm to forecast entrepreneurship population using Fuzzy Time Series

Abstract The development of this article required researchers to run several simulations with an FTS model, to project the entrepreneurial population under different values of the parameter C belonging to the bell-shaped membership function. Taking "entrepreneurial population growth" as a linguistic variable, researchers analyzed the potential of the model to adjust to the dynamics of the phenomenon under study, and used an FTS time-invariant. Algorithms were developed in R to run the different stages of the projection model; and experiments using rates as a measure to eliminate sample bias were carried out. The sensitivity of the model to different values of C ranging from 0 to 1 was also analyzed. The results show that the model is robust to project the historical data of the entrepreneurial population in Colombia. The sample data are unstable, however, when projecting the rates, the results are consistent. The lowest error is 0.04142 for C=0.01. Researchers also performed a comparative analysis to evaluate the potential of the FTS-R method to project the entrepreneurial population.