Browsing by Author "SUMIT CHOWDHURY"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
ThesisItem Open Access A Study of Fuzzy Time-series Models in Agriculture(IARI, INDIAN AGRICULTURAL STATISTICS RESEARCH INSTITUTE, 2012) SUMIT CHOWDHURY; Himadri Ghosh)The time-series modelling and forecasting investigate relations between present and past observations for the sequential set of measurement over time. The area has been widely studied and traditional forecasting are frequently conducted by various statistical tools. However, the classical time-series theory assumes values of the response variable to be „crisp‟ or „precise‟, which is quite often violated in reality, and cannot handle „fuzziness‟ in the underlying system, a problem which can be solved by fuzzy time-series analysis. Works had been done on various models for fuzzy time-series analysis. However, not much of a work is done till date on forecasting of out-of-sample data using fuzzy time-series models. In this paper, attempts have been made to develop a new methodology for fuzzy time-series modelling using non-convex membership functions and perform prediction of out-of-sample data. The performance of the methodology has been critically assessed by comparing it with the existing methodologies at various stages viz. modelling, validation and out-of-sample forecast.ThesisItem Open Access A Study of Fuzzy Time-series Models in Agriculture(IARI, INDIAN AGRICULTURAL STATISTICS RESEARCH INSTITUTE, 2012) SUMIT CHOWDHURY; Himadri GhoshThe time-series modelling and forecasting investigate relations between present and past observations for the sequential set of measurement over time. The area has been widely studied and traditional forecasting are frequently conducted by various statistical tools. However, the classical time-series theory assumes values of the response variable to be „crisp‟ or „precise‟, which is quite often violated in reality, and cannot handle „fuzziness‟ in the underlying system, a problem which can be solved by fuzzy time-series analysis. Works had been done on various models for fuzzy time-series analysis. However, not much of a work is done till date on forecasting of out-of-sample data using fuzzy time-series models. In this paper, attempts have been made to develop a new methodology for fuzzy time-series modelling using non-convex membership functions and perform prediction of out-of-sample data. The performance of the methodology has been critically assessed by comparing it with the existing methodologies at various stages viz. modelling, validation and out-of-sample forecast.