Forecasting ICT development through quantile confidence intervals

Regression is a common method to calculate relationships between variables. Quantile regression extends the calculation to the coefficients of various quantiles, providing a more complete overview. In addition, quantile forecasting models forecast coefficients. This study proposes a new algorithm to calculate the quantile confidence intervals of the in-sample data to forecast the coefficients of the out-of-sample data. The algorithm analyzes ICT data for 78 countries between 1999 and 2010. Results show that the algorithm provides valid forecasting results and outperforms previous studies. These quantile confidence intervals can also forecast the independent variables' impact trends on the dependent variable. The algorithm is applicable to different domains.