Forecasting High-Order Multifactor Cross Relations Using Fuzzy Time Series and Frequency Density

The forecasting technique frequently employed for decision-making, initially presented by Song and Cissom utilizing fuzzy logic principles, is Fuzzy Time Series (FTS). The FTS forecasting approach comprises fundamental steps: identifying the conversation universe, segmenting the conversation universe, fuzzification, constructing the fuzzy logic relationship (FLR), and defuzzification. This study involved the modification of the existing methodology to ascertain and categorize the conversational universe, fuzzy logic relationships (FLR), and historical data variations, each utilizing distinct frequency density partitions and cross-relationships, specifically with Indonesian rubber production data. Altering the frequency density partition algorithm to achieve partition universality, followed by fuzzification, cross-relationship analysis, and concluding with the defuzzification procedure. The degree of forecast precision uses the Average Forecasting Error Rate (AFER) to evaluate the revised outcomes against the traditional method. Simulations utilize Indonesian rubber production data from 2000 to 2023, exhibiting an AFER of 3.901%, below 10%, indicating robust forecasting criteria. The forecasting technique frequently employed for decision-making, initially presented by Song and Cissom through the principle of fuzzy logic, is Fuzzy Time Series (FTS). The FTS forecasting approach comprises fundamental steps: identifying the conversation universe, segmenting the conversation universe, fuzzification, constructing the fuzzy logic relationship (FLR), and defuzzification. This study involved the modification of the existing methodology to ascertain and categorize the conversational universe, fuzzy logic relationships, and historical data variations, each employing distinct frequency density partitions and cross-relationships, utilizing Indonesian rubber production data. Altering the frequency density partition algorithm to achieve partition universality, followed by fuzzification, cross-relationship analysis, and concluding with the defuzzification procedure. The degree of forecast precision uses the Average Forecasting Error Rate (AFER) to juxtapose the revised outcomes with the traditional methodology. Simulations utilize Indonesian rubber production data from 2000 to 2023, exhibiting an AFER of 3.901%, below 10%, indicating robust forecasting criteria.