MODELING TIME SERIES DATA USING FUZZY CLASSICAL AND FUZZY MARKOV CHAIN

• Main Author: Awal, Aidil
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/42309
• Institution: Institut Teknologi Bandung

One of the most widely developed forecasting methods is time series using a quantitative approach with past data. The forecasting process is very important in time series data because it is needed in the decision-making process. In the field of transportation, forecasting can be used to predict many traffic accidents. The development of time series data forecasting methods is quite rapid resulting in a large selection of methods that can be used to forecast data so it is necessary to compare one method with another method to get high accuracy forecasting results. The research describes the problem of modeling the forecasting of traffic accidents in the province of South Sulawesi using Fuzzy Time Series (FTS) developed by higher-order and Markov chain. The method development is done by improving the FTS method with mathematical rules and applied at the stage of the process of forecasting seasonal data on the number of traffic accidents. The test results show that the third-order FTS model has a higher forecasting accuracy value with the percentage calculation accuracy of Mean Square Deviation (MSD), Mean Absolute Deviation (MAD), and the best Mean Absolute Percentage Error (MAPE).