Minimal variability OWA operator combining ANFIS and fuzzy c-means for forecasting BSE index

Stock data sets usually consist of many varied components or multiple periods of stock prices, resulting in a tedious stock market prediction using such high dimensional data. To reduce data dimensions, it is crucial to fuse high dimensional data into a useful forecasting factor without losing information contained in the original variables. Decision makers may desire low variability associated with a chosen weighting vector, further complicating proper weight assignment for past stock prices. In this paper a new time series algorithm is proposed to overcome above mentioned shortcomings, which employs a minimal variation order weighted average (OWA) operator to aggregate values of high dimensional data into a single attribute. Based on the proposed model a hybrid network based fuzzy inference system combined with fuzzy c-means clustering is used to forecast Bombay Stock Exchange Index (BSE30).