PREDIKSI PENJUALAN BAHAN BAKAR MINYAK MENGGUNAKAN SISTEM INFERENSI FUZZY BERBASIS JARINGAN ADAPTIF ( ANFIS ) STUDI KASUS : PENJUALAN PREMIUM DI DEPOT PLUMPANG

• Main Author: Febrianto (NIM : 23298080), Firman
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/4912
• Institution: Institut Teknologi Bandung

<b>Abstract :</b><p align=\"justify\">A key component that support the operation of oil company is fuel demand forecasting. With proper forecasting, the whole planning activities of the company would be more effective and efficient. So, the accuracy of such forecasting has significant economic impact. <br /> <p align=\"justify\">The main goal of this research is to apply pattern recognition ability of ANFIS to the problem of gasoline sales forecasting in Depot Plumpang Jakarta based on past time series data and create a software of gasoline sales forecasting. To build the ANFIS architecture for forecasting, a training data set composed of data pairs represent desired input-output pairs of the target to be forecasted, is needed for learning process. Learning process needs a training data set to obtain the best parameters of ANFIS that has minimal error . There are some ways to minimize error, such as create a best mapping from sample data, change the number and type of membership function and number of epoch. The software of gasoline sales forecasting programmed based on the best parameters of ANFIS using C++. The second goal of the research is to analize the result of ANFIS in special days and update the error measure, analize cascade forecasting and compare the result of ANFIS with result of forecasting using autoregression method and artificial neural network to see the performance of ANFIS.