Best Method among three Trend Analysis Methods and Its Forecasting
Budi Pratikno *
Department of Mathematics, Faculty of Mathematics and Natural Science, Jenderal Soedirman University, Purwokerto, Indonesia, Jl. Dr. Soeparno 61 Karangwangkal, Purwokerto, Banyumas, 53122, Central of Java, Indonesia.
Anindya Gunawan
Department of Mathematics, Faculty of Mathematics and Natural Science, Jenderal Soedirman University, Purwokerto, Indonesia, Jl. Dr. Soeparno 61 Karangwangkal, Purwokerto, Banyumas, 53122, Central of Java, Indonesia.
*Author to whom correspondence should be addressed.
Abstract
We study the trend analysis methods, namely (1) least square method, (2) trend quadratics method, and (3) exponential trend method, to get which one is the best in forecast the human development index (HDI) time series data. Here, we used mean absolute percentage error (MAPE) in comparing the three methods to obtain the eligible one. A simulation is given then using on the HDI data from Cilacap regency. The result showed that the smallest MAPE (on the trend quadratics method) is 0.19%. Due to this value of the MAPE, we then choose the trend quadratics method as the best model. Furthermore, we used this method to forecast the simulation HDI data, and the result showed that the values of the forecasting are 70.6, 70.7, and 70.7 for period 2022 to 2024, respectively. These result indicate that the trend of the HDI in Cilacap tend to be slowly increase.
Keywords: HDI, forecasting, trend analysis, quadratic trend