Prediksi Pengguna Broadband di Indonesia sampai Tahun 2024 Menggunakan Pemodelan Crisp-DM dengan Metode Decision Tree dan Algoritma ID3 Studi Kasus di BPS dan APJII

Abstract

Indonesia is a country with a population of 270.20 million people recorded at the Central Statistics Agency in September 2020. Based on the International Telecommunication Union (ITU) reported, the United Nations (UN) body, the number of world internet users in 2018 was 3.9 billion, exceeding half the world's population. Today the position of the internet continues to be significant in social, economic, and political life in a globalized world. Every year the internet continues to influence human life. Therefore we need research on forecasting the number of broadband users in the future to find out the number of users will increase or decrease as well as planning for Indonesian broadband. This forecasting research uses, using a comparison, namely the double exponential smoothing method with two different alphas and two betas, then the single exponential smoothing method with two different alphas. The double exponential smoothing method with alpha 0.2 betas 0.4 with the result that the MAD, MSE, MAPE values are smaller than single exponential smoothing with the first part alpha 0.8 and the second part alpha 0.5 and smaller than the double method exponential smoothing with alpha 0.6 betas 0.2. After getting the best value from the double exponential smoothing forecasting method, the prediction process is continued using the decision tree method, namely the ID3 algorithm for decision making every year from 2000 to 2024 to expand bandwidth or by not expanding bandwidth with the result that there are 19 years to expand bandwidth and there are 6 years for the decision not to expand the bandwidth.

Keywords

• forecasting, prediction, exponential smoothing, decission tree method, id3 algorithm.