Replication Data and Code for: Forecasting Internet Penetration for Digital Transformation and Sustainable Development in Somalia: A Comparative Evaluation of Statistical, Neural Network, and Hybrid Forecasting Models
Description
This dataset and computational package contain the empirical data, pre-processing routines, and forecasting replication scripts supporting the study: "Forecasting Internet Penetration for Digital Transformation and Sustainable Development in Somalia: A Comparative Evaluation of Statistical, Neural Network, and Hybrid Forecasting Models". The research investigates the historical evolution and future diffusion of internet penetration (percentage of individuals using the internet) in Somalia over the period 1990–2022. The repository provides the raw multi-country indicator dataset, the extracted and calendar-reconstructed Somalia time-series matrix, and the complete replication scripts in both R and Python to reproduce all unit root tests, model estimations (ARIMA, ANN, ARNN, Naive persistence, and residual-based Hybrid ARIMA–ARNN), out-of-sample holdout validations (2017–2022), and medium-term projections (2023–2028).
Files
Institutions
- Amoud UniversityAwdal, Boorama