Fuzzy Time Series Model Cheng dengan Interval Fibonacci dalam Prakiraan Air Quality Index

Abstract

Air quality in Indonesia has become increasingly concerning due to the rise in urbanization and industrialization activities, making it essential to employ accurate forecasting methods to monitor the Air Quality Index (AQI). One commonly used method is the Fuzzy Time Series Cheng model; however, its accuracy can be affected by less adaptive interval partitioning. This study aims to improve the accuracy of AQI forecasting by applying the Fuzzy Time Series Cheng method using Fibonacci ratiobased intervals as an innovative approach to achieve more proportional and adaptive interval partitioning. The data used are secondary daily AQI values from cities in Indonesia, covering the period from 7 February 2025 to 8 March 2025, which were then categorized based on stationary and non-stationary data patterns. The results show that the use of Fibonacci ratios can improve forecasting accuracy for both data patterns, with the average Mean Absolute Percentage Error (MAPE) decreasing from 12.73% to 12.26% for stationary data and from 14.59% to 13.78% for non-stationary data. Therefore, Fibonacci intervals have been proven to be more effective in enhancing AQI forecasting performance using the Fuzzy Cheng method.