Forecasting industrial production index with artificial neural networks and fuzzy logic

Industry is the supply, processing and production of raw materials to make them suitable for human use. Today, one of the indicators showing the status of the industrial sector and monitoring the increase or decrease in production activity comparatively is the industrial production index. The industrial sector is an important precursor in determining the gross domestic product. The main reason for this is that industrial production affects the economy in many ways. Whether it is domestic consumption or exports, the increase in industrial production is related to economic growth. Since industrial production is sensitive to both consumption needs and interest rates, the industrial production index is directly related to growth. Since the industrial production index is one of the most important indicators of growth, it is also important to predict the future direction of the industrial production index. In this master's thesis, artificial neural network and fuzzy logic models have been produced by using data from previous years to predict the electricity production index for a year. Different models were obtained by changing the transfer functions and number of neurons in ANN and these models were used in forecasting. Similarly, for ANFIS models, diversification was obtained by changing the membership function and the number of membership functions. The predictions made in ANFIS were performed on these models. When ANN models are analysed, it is also observed that the system consisting of time, mining and manufacturing production indices inputs, as in ANFIS, gives better results than the system that forecasts only based on time or only manufacturing-mining production indices.